Related to—and sometimes synonymous with—agile manufacturing, adaptable manufacturing, cellular manufacturing, computer-integrated manufacturing, flexible machining, and flexible automation, flexible manufacturing encompasses a diverse set of manufacturing principles and technologies with a few common goals:
More specifically, flexible manufacturing systems (FMS) generally consist of a combination of computerized numerical control (CNC) machines, robots, self-diagnostic systems, and a hierarchical information system. They may also include any number of other devices for material handling and other functions. These systems are designed to be easily reprogrammed or even regrouped with other devices in order to permit rapid and inexpensive changes in the manufacturing process, facilitating quick responses to market changes and allowing for so-called mass customization of products. FMS can likewise monitor, schedule, and route workflow to maximize efficiency and to avoid deadlocks due to component failures or backlogs on certain devices.
Although each FMS is unique in configuration and process, an abstract description of overall operations in outline form is possible. First, numerical control programs and computer-aided process planning are utilized to develop the sequence of production steps for each machined part. Next, based on inventory, orders, and computer simulations of how an FMS can run most economically, a schedule is established for parts that are going to be produced on that day. Following that, material and required tooling are retrieved either automatically or manually from storage and loaded into the system. Once loaded, the FMS begins machining operations. According to the process plan, robots, conveyors, and other automated material handling devices transport the workpiece between workstations. Should any tool break down during the production cycle, an FMS can reroute the workpiece to other tools within the system.
Most FMS (or their subsystems) are designed to produce any number of related workpieces. In general, the essential characteristics that constitute a workable "part family" are common shape, size, weight, and tolerance range. Since members of a part family share many traits, usually a software-directed change is all an FMS requires to switch from producing one kind of part to another.
Workstations, or individual processing units, are a concept central to any discussion of FMS. A processing unit refers to the sum of operations performed within a particular factory area containing several pieces of process equipment meant to carry out individual or multiple operations on various products. Indeed, different workstation distinctions mark the divide between two FMS subsystems—the flexible manufacturing module (FMM) and the flexible manufacturing cell (FMC). The module comprises a stand-alone numerically controlled machine tool (NCMT), automatic material handling device (such as a robot or automatic pallet changer) and an automated monitoring system to control for tool breakage, equipment depreciation, automatic measuring and related diagnostics. FMM constitute the first step in the automation of the loading and unloading of parts to and from an NCMT. For machining centers, automatic pallet changers make it possible for unmanned attended machining to occur for up to six hours.
The flexible manufacturing cell, on the other hand, comprises two or more machine tools which may or may not include NCMT. Similar to FMM, the FMC incorporates a material handling device (such as a robot) servicing several machine tools arranged in a circle or line. Automatic pallet changers are also used in conjunction with automatic conveyor systems linking NCMT. Compared to an FMM, information flow to and from an FMC is integrated into a larger monitoring control system. FMC are less flexible than FMM. Usually FMC are applied to a "family," as opposed to some broader grouping of components—for instance shafts within a prescribed size range. Because each FMC is designed to meet the specialized requirements of different customers, standardization is not a universal feature. In fact just the design phase requires considerable consultation and information exchange between the specific user and FMC supplier. It should be noted that early versions of FMC and FMS initially functioned below their anticipated performance level. Problems arose due to the technical difficulties involved when linking product flows with different machines. As a result, demand for FMC is initially limited to a few large firms with enough financial clout to undertake the risky investment the systems represented.
With the above subsystems in mind, FMS can be distinguished by the following characteristics: the flow of tools and parts between different machine groups is automated; material handling is mostly, but not exclusively, performed by automated guided vehicle systems (AGVS), and not, as in the case of FMC, industrial robots. This is explained by the fact that FMS consist of machining centers mostly involved with the production of, geometrically speaking, prismatic parts, while material handling robots work in conjunction with computer numerically controlled (CNC) lathes that machine rotational parts. To a far greater extent FMS also incorporate the use of conveyors and rail guided transport systems. And, in comparison with their subsystems, FMS have a lower rate of labor utilization, higher labor productivity, and, in certain cases, higher capital productivity. FMS are usually equipped with a "hot standby" feature. This alludes to a computer control system comprising two computer units. Should one fail to operate, the other automatically kicks in to keep the FMS running.
In terms of joint characteristics, FMS and their subsystems share most, if not all, of the following to differing degrees:
Since all flexible systems are not equally versatile in all respects, manufacturers must determine which aspects of flexibility best serve business objectives such as cost efficiency, speed, or the ability to rapidly implement product alterations. These objectives will dictate what kinds of tasks are best suited for flexible solutions and how much capital should be invested to develop each component of the FMS.
Besides true FMS and their subsystems, are two other manufacturing concepts distinguished by their comparatively higher level of automation and lower level of flexibility. These are the flexible transfer line and fixed transfer line. Between these two systems and FMS a trade-off, or "productivity dilemma," is said to exist. Accordingly, the higher the level of automation, the greater the manufacturer efficiency in terms of productivity and lower unit costs. But the efficiency advantage held by non-FMS related systems is said to be achieved at the expense of losing a certain degree of innovative flexibility.
The flexible transfer line comprises workstations utilizing numerous automated general or special purpose machine tools joined by an automated workpiece flow system according to line principle. The flexible transfer line is capable of simultaneously or sequentially machining different workpieces running along the same path. A fixed transfer line, by contrast, utilizes a number of special purpose machine tools (as opposed to general ones) initially designed to produce one product only. After a considerable period of setup time though, it can accommodate a different variant of the product, as in the case of different sized cylinder heads.
From an economic standpoint, if the number of variants is high (500 or more variants) but produced in relatively small annual amounts, the use of stand-alone NCMT (a portion of which included FMM) is considered appropriate. Progressing to larger annual production of fewer variants, FMC, followed by FMS, become optimal strategies. FMS then, occupy a middle ground between highly automated and highly flexible manufacturing systems. Based on several empirical studies conducted in the 1980s, the respective variants range for FMS stands between 4 to 100; for FMC, between 40 and 800. For workpieces of different types ranging from 2 to 8, flexible transfer lines are used, and for I or 2, fixed transfer lines.
More than a few manufacturing experts have taken issue with the so called "productivity dilemma" as described above, countering that it is at best a transitory trend. They look instead to the further development of computer-aided manufacturing (CAM) as the potential solution for reversing the short term opposition between cost efficiency and inflexible production units. Computer-based technology, they have argued, reverses the trend towards specialized hardware and replaces it with specialized software. Thus general purpose machines utilized in flexible transfer lines can be programmed with special purpose software to achieve the flexibility once attributed to an earlier but narrowly conceived definition of a FMS. Simply by reprogramming a general purpose machine, the switch to a new product variant can be accomplished without having to rip out and replace the machine, as was the case in earlier times. Also unlike times past, this task can be completed in a matter of hours instead of weeks.
One of the outstanding features of FMS is computer simulation. Thirty years after their introduction, a plethora of microcomputer-based simulation packages are available for simulation analysis. There are also many mainframe simulation packages now available in a microcomputer format. Unlike their mainframe forerunners, microcomputer simulation packages are considered user-friendly and do not require the presence of highly skilled individuals. Subsequent developments introduced sophisticated color graphics in conjunction with several advances in computer mainframes that allowed software developers to create even more complex programs that retain their user-friendly orientation. A noteworthy advance is the incorporation of animation techniques. In the opinion of Nigel Greenwood, had simulation analysis been available in the early days of FMS, their success rate would have been greatly enhanced.
Computer simulation exercises are performed to identify system bottlenecks—the utilization of fixed resources such as machine tools and variable resources such as operators, tools, and material transporters. Utilization of resources is measured in terms of busy, idle, down, and blocked time. Simulation results then provide information to FMS designers about work in progress, production rates, and the impact of equipment failure, all relevant factors that determine how well a particular FMS works. When computer simulation packages first appeared, many experienced production engineers questioned their usefulness. Having been schooled in a "hard modeling" simulation background that used scale layout, pieces of paper, drawing pins, and the like to represent personnel and pallets, experienced engineers were skeptical of computer simulation modeling. The substantial advantages of computer simulation and the influx of new engineers with computer trained backgrounds has alleviated this initial skepticism.
One critical limitation of computer simulation is that computer simulation modelers are unable to accurately depict a specifically designed FMS facility, so that the possibility of an informational discrepancy between FMS project designers and simulators exists. Heightened communications between these two groups ensures that discrepancies are minimized. It is also expected that further developments in microcomputer-based packages will help minimize discrepancies.
In the evolution of an FMS during its multiphase design process, computer simulation is of great assistance. During the conceptual design phase, if given a hypothetical array of workstation configurations, a computer simulation is able to determine their respective throughput times to a high degree of reliability. It is also used to calculate initial financial analysis and to animate prospective configurations. During the detailed design phase, computer simulation makes it possible for suggested system improvements to be incorporated and tested in an FMS model to analyze their impact.
The principal motivation for developing and implementing an FMS is to ensure that the transformation of processed and unprocessed raw materials to finished parts is as rapid, efficient, and well-conceived as possible. Of the three elementary FMS processes, material handling devices and computer control systems offer the greatest potential for improved performance results. The other element of FMS—the manufacturing process—has experienced significant progress in terms of speed, reliability, and efficiency, which led to a sizable reduction in overall product throughput time. In turn, a greater emphasis on the implementation of automated material handling systems to increase the overall efficiency of FMS is paramount. It has been estimated that close to one-third of an FMS product's total manufacturing cost is absorbed by the expense of successive material handling tasks.
FMS incorporate four major types of material transport tasks: transport between different systems; transport between different subsystems within the same FMS; transfers between workstations within various subsystems; and transfers within the workstations themselves. From the standpoint of an FMS designer, it would be ideal to address these tasks in their totality, but the incompatible variety of the tasks or load types makes this impossible.
Given their relatively versatile property and high load capacity, automated guided vehicle systems figure prominently in most FMS. They are battery-driven, bidirectional vehicles designed to automatically transport material from one location to another along a predefined route or set of routes. Programming is accomplished using microprocessor controls or wire guided systems. Of the two, wire guided vehicles are preferred. Transport instructions from a central computer are received via a wire buried an inch below the shop floor through inductive sensors located on either side of the vehicle. Loading and unloading operations are performed by hydraulic lifts or in a manner similar to traditional fork-lifts.
Considered less flexible and unable to duplicate the high load capacity of an AGVS, conveyors are thought to be just as reliable, relatively less expensive, and battery-free. Conveyors are useful for frequent transportation tasks. Because of their mechanical simplicity, conveyors can transfer loads efficiently using a large number of sensors connected to a conveyor control system. Conveyors are available in a wide variety of sizes, speed capabilities, and forms. Among some of the more common industrial types are overhead monorail, carry and free, power and free, under-floor drag chain, floor slat, gravity feed, and plastic chain link.
Rail guided transport represents a blend of conveyor and AGVS systems. Instead of being guided by a underground wire, above-ground rails determine routing direction. Compared to an AGVS, rail guided transport moves more rapidly along straight distances that typically range from 30-50 meters, but do not have as much route flexibility. Because of their relatively unsophisticated control systems and battery-free operation, rail guided transfer systems have many of the advantages attributed to conveyors.
Within FMS subsystems, robot and gantry loaders predominate. Robots are mostly used for tasks directly related to the manufacturing process rather than material handling. Under the appropriate circumstances, they can be used for material transport, as their inherent flexibility often offsets their relatively high cost. Programmable logic controller gantry loaders combine the advantages of robots with conveyors. Indeed many of these are simply an overhead rail-mounted robot or a robot arm moving along a rail. This gives them a speed advantage over robots, but being mounted on a fix path (albeit with several stopping points) reduces their flexibility. Manual material loading or fork-lift operations are the most inexpensive material handling and most versatile form of transport. The advantage held by automated systems however, is their continuous operation and accurate and consistent performance.
Computer control systems account for the largest single risk factor within FMS. Between 50 and 75 percent of the total risk involved in FMS implementation is attributed to the control system, which includes both the computer hardware and associated software. In overall cost terms, computer hardware ranges from 10 to 15 percent while its associated software represents 15-30 percent.
This high risk factor is associated with several problems, the most pronounced of which is that the "mechanics" of many FMS are designed by machine-tool inclined engineers, while the software is designed by persons who have little background in the manufacturing environment they are attempting to create products for. An additional problem is that FMS software is extremely difficult to evaluate prior to being installed and tested. If found to be inadequate, corrective steps are expensive.
The principal objectives of an FMS computer control system are fourfold: to facilitate the transmission of support software to material handling systems and manufacturing process equipment; to coordinate the material handling system to allow the manufacturing process equipment to function at it highest level of utilization; to facilitate data entry, control, operation, and monitoring of the FMS as a whole; and to guide the return of the system to its complete operational status following a failure.
At the turn of the 20th century, one of the most active areas of FMS development is directed at producing a "generic" computer control system to replace customized systems. While the prospects for success are mixed, incremental progress is being made in reducing the costs and risks associated with software developments.
In the early 1970s, after a period of initial euphoria, disappointment set in when FMS failed to live up to their potential. Since the early 1980s, however, FMS proved to be technically feasible, reliable, and cost-effective. Indeed, by the 1990s, it was widely accepted that production principles once thought to be applicable only at the level of the small job-shop (a small manufacturing facility typically engaged in making a small batch of a wide variety of different parts for a large customer) were equally successful in mass production operations.
The Wealth of Nations, Adam Smith's famous analysis of the division of labor and its impact on specialization, productivity, automation, and product flow, marked one of the first recorded commentaries on changing production techniques in the early modern era. Nearly 130 years later, thanks to Henry Ford, the assembly line method of mass production represented the perfect illustration of Smith's ideas. Yet because Ford's highly automated assembly line operated with only the slightest degree of product variation, it represented the virtual antithesis of flexible manufacturing. Between the two World Wars, the automated assembly line dominated the industrial sector because of its unsurpassed efficiency and productivity.
Two developments following World War II proved critical in the development of flexible manufacturing systems. These are the manufacturing philosophy of group technology and the introduction of numerical control machine tools (soon to be followed by the related development of computer numerically controlled—CNC—machine tools and other applications). Both developments have been instrumental in narrowing the gap between the increased variety characteristic of FMS and the efficiency associated with assembly line automation.
Group technology was based on the principle of using the common properties of similar parts to rectify shared design and manufacturing problems. This involved the creation of a coding system to classify parts into groups given their shape, overall dimension, required accuracy, surface finish, manufacturing requirements, and material composition. Afterwards, an analysis is performed to determine all the major components a firm might manufacture. The acquired information is then sorted into major categories, such as processes or products, to indicate common production groupings in order to determine the appropriate set of machine tools necessary for their manufacture.
Initially, because the set of machine tools used for producing these parts was kept to a minimum, a small scale manufacturing shop was able to produce the parts more efficiently than a larger manufacturing unit. The connection between group technology and FMS stems from the sorting process that occurred within group technology. Sorting parts into well defined groups, performing an analysis, selecting the parts, and determining the most appropriate machine grouping—all of these steps are necessary to derive the maximum benefits from a FMS investment.
The cost-saving impact of group technology was illustrated in a 1986 study. Components produced by one large U.S. manufacturer required over 100 operations and had a throughput time (a measure of a component's processing time from start to finish) averaging five months, during which time they traveled some three miles on the shop floor. Following the implementation of group technology, the number of operations decreased from 100 to 10, throughput time fell to just days, and the distance traveled on the shop floor decreased to 200 feet. A factor critical to the success of group technology was the willingness of the workforce to acquire a multiskilled background that reduced the number of shop-floor skill levels from 11 to 2.
Numerical control machine tools, on the other hand, first emerged in 1952. Parallel developments were undertaken at the Massachusetts Institute of Technology and by Alfred Herbert in the United Kingdom. Initially, these continuous-path machines proved unreliable and it was 1956 before more reliable machines were developed.
It is largely the aerospace industry that encouraged the development of NCMT because it frequently required "copied" components that could only be produced repetitively using relatively inaccurate hand made templates. The computer equipment used to control machines engaged in manufacturing aerospace components was at first cumbersome and expensive, but advancements in computer technology in the area of complex integrated circuits ushered in the era of computer controlled machine tools. Compared to their numerical control predecessors, small minicomputers had far more versatile and reliable control devices. They were less expensive, incorporated computer-aided programming, and featured sophisticated editing components. These benefits meant machine tools were more flexible, more programmable, and easier to set-up when producing a variety of components. Developments in computer numerically controlled machine tools culminated in machining centers, which incorporated automatic tool changers, tool storage systems, automatic work transfer, and off-line programming capabilities.
The development of CNC machine tools provided the major technological foundation upon which early FMS versions were built. The tools made it possible for a computer, instead of a skilled operator, to control a manufacturing process. At first computer numerical control was applied only to machine tools. Later it is broadened to a wide variety of manufacturing processes. Further developments in CNC applications led to the creation of advanced industrial robots. Among other tasks, robots are used for material handling purposes, arc and spot welding, and paint spraying. The combination of CNC technologies and advances in computer technology allowed FMS designers to introduce such breakthrough tools as programmable logic controllers (PLACE), automated guided vehicle systems (AGVS), and automated storage and retrieval systems (ASKS). Yet despite these advancements, FMS encountered difficulties when it came time to smoothly integrate these major elements into a dependable working system while minimizing the costs and risks attributed to once-off control software.
The first FMS (albeit rudimentary) was installed in England in 1968. Some authorities dispute this however, arguing that Project Tinkertoy, carried out under the auspices of the U.S. National Bureau of Standards in 1955, represented the first FMS. The English FMS was developed by D.T.N. Williamson of the Molins Machine Tool Company; it was known as System 24. The confluence of several factors prompted its development: the advancements made in machine tool technologies, theoretical progress surrounding the area of job-shop scheduling, the growing influence of group technology in different manufacturing environments, and the increasing demands placed upon larger manufacturing establishments to produce components economically in small batches.
Similar to later FMS developments, System 24 was designed to feature three specific qualities: the ability to produce a large variety of components virtually at random; the ability to both load and unload tools and workpieces automatically; and the ability to operate without manned attendance over an extended period. The eventual demise of System 24 came about due to the system's unreliability and the fact that the light flat alloy components market for which it was developed did not expand as had been predicted. As a result, the development of FMS techniques in England was abandoned for nearly ten years; FMS research and development pressed onward in the United States and Japan.
In their early phase, many FMS encountered a common set of difficulties that were eventually corrected. First, process equipment initially proved to be insufficiently flexible to meet varied production requirements. Second, many supporting technologies also proved inadequate. For instance, sensors built into early systems were usually limit switches. Experience had demonstrated that limit switches are highly unreliable, yet they were included in FMS in such a manner that their failure prompted the entire FMS to cease operation.
Additionally, most early FMS were overly machining-center oriented. This bias resulted from the fact that advances in machining centers outdistanced those of other machine tool types. Systems typically associated with cubic component workpieces progressed the quickest in design and implementation, computer control systems, and pallet and tool changers; systems associated with rotational workpieces lagged considerably behind.
Because they can dramatically change the manufacturing process when implemented, FMS can also have significant impact on a company's labor and organizational structure. Three main issues are involved: (1) workplace organization; (2) their significant employment/labor-saving component at the company level; and (3) changes in skills needed to run FMS versus traditional functional manufacturing processes.
Although there have been mixed findings in the academic literature, the conventional wisdom is that flexible systems lend themselves to decentralized organizational structures. This is allegedly because they are information-intensive but require comparatively fewer people with broader skills to run them. Studies suggest that the effects on organizations from installing flexible automation may be much less than once believed; while the two phenomena may be correlated, there is not much evidence that flexible automation actually causes decentralization. Instead, it may be that the same forces that tend to make companies pursue flatter management structures also encourage the adoption of flexible manufacturing practices. Nevertheless, because the information channels and labor requirements are different under FMS, organizational design is a valid concern for companies using them. Indeed, some evidence suggests that, in contrast to the decentralization theory, FMS can lead to decision making being further removed from the individuals who run the systems.
Two persistent, parallel, and interconnected economic forces figured largely in the development of FMS, which in turn prompted a substantial reorganization of the workplace environment. First, competitive pressures to lower unit costs are typically achieved by boosting productivity through the introduction and diffusion of technological change. This applied just as readily to FMS production techniques as it did to other groundbreaking techniques. And, as is true in previous times, active periods of technological change usually entail a significant de-skilling component.
Second, in terms of changes in market demand, a heightened presence of market uncertainty (rapid demand surges followed by just as rapid declines) stimulated FMS development. The impact of increased market uncertainty required that workers respond in a manner consistent with changing demand conditions, which in turn gave rise to the notion of labor flexibility. With respect to the organization of the workplace, this notion contained three potential policy scenarios related to flexibility: functional, numerical, and financial.
Functional meant the ease with which tasks performed by workers could be adjusted to meet changes in technology, markets, or company policy. Functional flexibility further grouped skill backgrounds into core, dual, and multiskill classifications. Core skill referred to cases where the practice of an existing craft or trade continued but, as the need arose, performance of unrelated tasks was required. Dual skill meant retention of the existing skill but also the requirement that the worker become proficient at a second task, for instance a machine operator undergoing training in hydraulic systems in order to maintain industrial robots. Multiskill referred to workers who had obtained skill proficiency in three or more tasks.
Numerical flexibility referred to the ease with which the number of workers could be adjusted to meet demand fluctuations using alternative types of employment contracts. These included annual hours, various forms of job sharing, part-time, and minimum/maximum hours working arrangements.
Financial flexibility captured the degree to which the structure of pay encouraged and supported functional and numerical flexibility. By the 1990s, while only a handful of FMS firms had incorporated the full complement of labor flexibility components as describe above, it was widely expected that, in time, they were to become the norm. Several case studies suggested that functional flexibility was more likely to be found in capital-intensive FMS firms, while numerical and financial flexibility practices were growing among relatively more labor-intensive ones.
As concerned the de-skilling impact of FMS, Edquist and Jacobsson (1988) conducted research in Sweden that indicated that the diffusion and displacement of conventional machine tools by NCMT led to a significant reduction in the skills per worker needed to operate NCMT, as well as in the number of people required to learn these skills. They reported that the maximum time required for an unskilled person with a technically oriented secondary education to become proficient with a NCMT ranged from 6 to 12 months. For skilled operators using conventional machines, by contrast, five years experience was often necessary to acquire proficiency. Productivity gains ranging from 50 to 60 percent were reported by substituting semiskilled computerized numerically controlled (CNC) machine operators for skilled conventional operators. Increases were reported, however, in newly skilled NCMT and CNC support workers, such as setters, programmers, and repair and maintenance engineers. All of these positions emerged with the spread of nonconventional machine tools and were not previously required as support labor to operate conventional machines. But in the long run, the implementation of flexible systems is expected to continue the decades old trend of declining employment in manufacturing trades.
Since flexible systems can represent a threat, or at least a challenge, to traditional manufacturing workers, it is important that management address worker issues early and effectively. Securing cooperation from workers is not only necessary to maintain employee motivation and morale, but it may also hold the key to the success of the new system. In one noteworthy comparison, a highly automated General Motors plant delivered lower productivity than another of the automaker's less automated joint venture plants. Research suggests that, more than anything, training and preparation for the new environment is most beneficial to gaining employee buy-in. Training, or perhaps re-skilling, was found to be more important than financial incentives or even job security.
[ Daniel E. King ]
"Adaptable Manufacturing: Agility to Cost-Effectively Produce on Demand." Modern Materials Handling, 15 May 1998.
Anderson, David M. Agile Product Development for Mass Customization. New York: McGraw-Hill, 1997.
Chen, Injazz, Atul Gupta, and Chen-Hua Chung. "Employee Commitment to the Implementation of Flexible Manufacturing Systems." International Journal of Operations & Production Management, July 1996.
Edquist, Charles, and Staffan Jacobsson. Flexible Automation: The Global Diffusion of New Technology in the Engineering Industry. Oxford, UK: Blackwell Publishers, 1988.
Jha, Nand, K., ed. Handbook of Flexible Manufacturing Systems. San Diego: Academic Press, 1991.
Oleson, John D. Pathways to Agility. New York: John Wiley & Sons, 1998.
Tidd, Joseph. Flexible Manufacturing Technologies and International Competitiveness. London: Pinter Publishers, 1991.
Tsubone, Hitoshi, and Mitsuyoshi Horikawa. "Impact of Various Flexibility Types in a Hybrid Fabrication/Assembly Production System." International Journal of Production Economics, 20 April 1999.
Vajpayee, S. Kant. Principles of Computer-Integrated Manufacturing. Englewood Cliffs, NJ: Prentice Hall, 1995.
Vokurka, Robert J., and Gene Fliedner. "The Journey Toward Agility." Industrial Management & Data Systems, March-April 1998.
Ware, Richard S. "Flexible Automation Reconfigures the Future." Machine Design, 21 March 1996.