Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
APPENDIX A The Technology of Future Manufacturing Although ah the interrelationships and long-term implica- tions of advanced manufacturing technologies are not yet wed un- derstood, the direction of future developments is relatively clear. This appendix describes the technologies that are likely to have a major impact on manufacturing competitiveness, indicates the ways in which those technologies interact, identifies additional re- search needs, and discusses some of the issues that are likely to be encountered in implementation. The technologies have been divided into materials, material handling, material transforma- tion, and data communication and systems integration. These categories are highly interdependent, and divisions are not always distinct, but this categorization provides an effective structure for a broad overview of the major technologies. DEVELOPMENTS IN MANUFACTURING MATERIALS Materials developments have a substantial impact on manu- facturing in both product design and process engineering. New products can require different materials and materials process- ing, and new materials themselves often spur new products and new process development. Many of the materials developments in manufacturing do not involve new materials, but rather substitu- tions, upgradings, and new concepts for conventional materials. Pressure to lower costs and raise product quality has led to some 75
76 major shifts in materials selection. The consideration of doubly precoated steel for corrosion resistance in automobiles is an ex- ample. Developments in both conventional materials and new materials are equally significant to future manufacturing. This section will focus on materials developments that are ready for manufacturing implementation, with minimal emphasis on research systems that have had little technology transfer effort. Developments in metals, polymers, ceramics, and glasses will be considered, followed by a brief discussion of some emerging issues that should be brought to the attention of policymakers. Metals and Metal-Based Composites Major developments will continue in the processing of con- ventional metallurgical systems. For example, large tonnages of carbon and stainless steed sheet and strip will be continuously cast. Similar developments are certain in nonferrous alloy areas as well. While the continuously cast products will have some mi- nor metallurgical variations to be considered, the major impact will be economic, allowing the basic metals industry to remain competitive in many sheet and strip applications. Increased use of warm- and cold-formed steel parts can be ex- pected, with emphasis on near-net-shape processes to save metal and avoid intermediate processing steps. Similar forces will con- tinue to drive powder metallurgical processing, although it must be emphasized that advances in precision forming and powder metallurgy have been slow over many years rather than a sudden breakthrough. Powder processing will be facilitated as cleaner powders reach the market. Powder metallurgy produced too! steels continue to offer advantages over conventional tooling stock. Superplastic forming will continue to increase in aerospace manufacturing; some initial applications have occurred in the B-1 bomber program. Increased market penetration will entail ma- jor changes in tooling and manufacturing technology. Thermo- mechanically processed 7000 series aluminum alloys are available for such forming, as are aluminum-lithium alloys. However, little of these alloys are available from domestic sources; most of this material is being obtained from the United Kingdom.
77 Many metal-processing alternatives will be examined to facil- itate in-line processing systems. Improvements can be expected in the control of metal structure, with increasing awareness of the importance of grain orientation (texture), residual stress, and sur- face quality. In steed surface treatment, increased use of induction or laser hardening can be expected as lower cost alternatives to carburizing. I.aser welding is seen as growing in the auto industry, perhaps at the expense of electron beam techniques. Power sys- tems manufacturing may turn to welded rotor fabrication to allow increased use of attractive alloy combinations while sidestepping large forging development problems. Many cases of metals substitution can be expected. The upgrading of alloys in small parts should increase product qual- ity, reliability, and processing response without grossly increasing overall metal procurement cost. Aluminum can be expected to continue to replace copper in many heat exchange applications. Silicon-based switches are expected to replace iron-based mag- netic devices. Also in the electronics industry, changes in plat- ing metals are expected, with gold giving way to palIadium-nicke! and iron. Some observers see increased use of molybdenum-based alloys, particularly as new developments solve some of the tradi- tional corrosion problems. Production of these alloys is energy intensive, however, and the domestic supply is limited. In the new metals area, continued progress is expected in the development of metal matrix composites, particularly using metals such as aluminum and magnesium reinforced with silicon carbide. Three major types of reinforcement are receiving particular at- tention: continuous monofilament, discontinuous, and continuous multifilament yarn. Each reinforcement requires a specific fab- rication process, including diffusion bonding, hot molding, power blending, forging, casting, pultrusion, extrusion, and rolling, often in combination. Many of the material defects and anomalies that have plagued metal matrix development are attributable to the manufacturing process. As experience in these processes builds and the manufacturing technology evolves, application problems caused by material defects should decrease. In aerospace struc- tures, for example, the application of specific metal matrix materi- als varies according to expected environment, design Toads, stress,
78 and temperature variations. Designers of flaw-critical parts select materials for their strength and ductility as well as their resistance to crack growth. Unfortunately, research on fracture mechanics in the area of metal matrix composites is much less developed than for polymer matrixes and has not been widely disseminated for use by designers.) Progress in reducing material defects and continued applica- tion experience will result in broader applications for metal ma- trix composites. For example, automobile manufacturers are gain- ing experience with aluminum-silicon carbide in piston ring and crankshaft applications. Power systems applications are seen for nickel superalloy and stainless steel matrix composites strength- ened with silicon carbide. Such systems allow increases in elastic modulus and desirable decreases in the coefficient of thermal ex panslon. Much research has focused on rapidly solidified metals and amorphous metals. Introduction of such materials into the manu- facturing sector is expected to be slow, with the major exception of the iron-boron-silicon-carbon system being broadly used for dis- tribution transformers. Slow emergence is also predicted for nickel and titanium aluminides despite extensive research. Recent alu- minide development has greatly increased its durability, and some jet engine applications can be expected. Polymers and Polymer-Based Composites Polymers and polymer-based composites will probably con- tinue to displace carbon steel and aluminum in a significant seg- ment of structural and paneling applications. This trend may be most prominent in the automobile industry, but it is also likely in electronic hardware, appliance chassis applications, and home building components. Current automobiles contain about 157 pounds of plastics and polymer-based composites. By 1995, this should grow to 213 pounds as polymer materials are used increas- ingly in body panels. The increased use of plastic is expected to save motorists about $200 per year through fuel economy, corro- sion resistance, cheaper repairs, and lower insurance rates.2
79 The applications seen in models such as the Pontiac Fiero will spread for reasons of economy and safety. In the Fiero, the hori- zontal body panels are made of flexible glass fiber-fi~led polyester (a sheet molding compound) and the vertical panels are made of relatively stiffer glass-reinforced polyurethane (a reaction injec- tion molded product). Beyond the body pane} substitutions, fur- ther use of polymers can be expected in the automobile structure. For example, polymeric leaf springs have been used in some auto- mobiles since 1981. While glass-polyester and glass-polyurethane materials will see major tonnage applications, future use of poly- mers in automobiles con be expected to rely on reaction injection molded thermoplastics as well. Beyond these applications, the ex- panded use of coatings (including paint) on steel can be regarded as an area where polymers will intrude further into the sheet metal markets. The growth of polymer panels and structural shapes has ne- cessitated the development of adhesives as a joining medium. In addition to the increased use of conventional adhesive materi- als, advanced work is under way on adhesive systems for higher- temperature applications (epoxies, polyimides), adhesives with greater strength and elastic range than epoxies, primerIess adhe- sives (silicones perhaps), and faster-curing adhesives (cyanoacry- lates, urethanes, etch. In some instances, adhesive development overlaps sealant systems (silicones). Beyond the relatively sim- ple polymer applications, adhesives are increasingly required for bonding dissimilar materials, particularly when differential ther- mal expansion must be accommodated. For example, rivets can- not be used to join plastic liners to metal trailer bodies because of differential thermal expansion. In some metal joining develop- ments, adhesives are being used in conjunction with spot welding to replace riveting. However, it must be emphasized that the use of adhesives for nonpolymeric joining has been slowed consider- ably by concerns about reliability, contamination, degradation, and consumer acceptance. Another major shift in polymer materials use can be fore- seen where flame retardation is a dominant consideration. Under- writers Laboratory interpretations of smoke, flame, and toxicity
80 requirements are leading to shifts away from polyvinyl chioride- based systems to fluorocarbons. High-Technology Ceramics Advanced high-technology ceramics3 are nonmetallic materi- als having combinations of fine-scale microstructures, purity, com- plex crystal structures, and precisely controlled additives. In con- trast to traditional ceramics, which are made from natural raw materials such as silica and clay, advanced ceramics are made from artificial raw materials, such as aluminum oxide, zirconia, yttria, silicon nitride, and silicon carbide, which are formed, sintered, and treated under precisely controlled conditions. The advantage of such fine ceramics is their ability to play both functional and structural roles. Functional uses include optical devices, motors, transducers, sensors, and semiconductors; structural uses include those that require high specific strength, high wear resistance, and high corrosion resistance. Both roles will be increasingly impor- tant in manufacturing applications. Currently, high-technology ceramics are used most often in electronics, including optical fibers, multilayer ceram~c-to-metal interconnecting and mounting packages for integrated circuits, ce- ramic multilayer chip capacitors, piezoelectric ceramic transduc- ers, and chemical, mechanical, and thermal sensors. Processing for these applications is generally an extension of standard ceramic technology, in which powders are pressed or formed with binders and sintered to density the ceramics. Incremental progress has im- proved results, but major improvements cannot be expected until semiconductor processing techniques are applied to ceramic com- ponents. Techniques such as selective-area ion implantation and laser-induced recrystallization will greatly improve many of the properties of electronics ceramics. High-technology structural ceramics are used as coatings and for monolithic and composite components. Major applications in- clude tooling for metal working, wear components in a variety of abrasive environments, bioceram~cs for bone replacement, and military ceramics for radomes and armor. Major efforts are un- der way in both the United States and Japan to use structural
81 ceramics in a variety of automotive applications, including engine wear components, turbochargers, bearings, and a variety of diesel engine components. Significant strides have been made in the mechanical prop- erties and reliability of monolithic structural ceramics. Under- standing of strength-limiting flaws and temperature-brittle frac- ture behavior has improved greatly, but further work is needed to improve reliability. Improvements are needed in powder synthe- sis, powder properties, near-net-shape fabrication methods, mi- crostructure control, mechanical properties, and nondestructive testing methods. Important research also is needed to identify new, more complex ceramics. Significant advances also are being made in thermal barrier coatings and ceramic matrix composites. Ceramic matrices com- bined with particulates, whiskers, or fibers of a different ceramic compound or metal have yielded composites with five times the resistance to fracture of the monolithic ceramics. Recent success has been reported in the use of metal ion implantation to reduce the relative friction resistance of ceramic diesel engine parts.4 New research is needed to quantify the improved mechanical properties of composites, particularly fracture resistance. In addition to the research required on the composition and properties of ceramic materials, much work is needed on the pro- cessing and product design requirements of ceramics. Promising directions in ceramic processing include the use of ultrafine pow- ders and the use of chemical routes to supplement or bypass some of the powder-processing stages. Other requirements include the processing of fine-scale layered structures, processing of ceramic composites, joining of ceramic parts, and near-net-shape process- ing of complex parts to minimize machining requirements. The rate of technical progress in ceramic materials and pro- cessing will determine the pace of commercial application. Major market penetration for structural ceramics depends largely on the progress made in automobile applications. Several Japanese firms have already introduced ceramic turbocharger rotors, piston rings, swirl chambers, and camshafts.5 An almost totally ceramic engine is a major research objective of virtually every automobile manu- facturer and should be extant by the early 1990s. As these auto
82 motive applications increase and the price falls, high-technology ceramics can be expected to see rapid growth in product and pro- cess (cutting tools) applications.6 General Issues Related to Materials It is important to recognize the critical lack of data on and basic understanding of the physical properties of many materials. This lack is a severe handicap in manufacturing process develop- ment. Most materials handbook data have been generated for use in product design and service performance analysis rather than for process analysis. The lack of knowledge has become acute as software systems have emerged with powerful process control and process design capabilities. The requisite data inputs often involve combinations of stress, strain, strain rate, temperature, heat transfer, friction, and so on, which have not been studied even for classic engineering materials. This lack of data on manufacturing materials grossly compro- mises the effectiveness of computer-based process models, and it tends to foster undue reliance on the few materials for which an adequate data base seems to exist. Power systems in particular are plagued by a lack of materials innovation due to the awesome data base requirements for service performance analysis, manu- facturing modeling, and code adherence. With the current low return on investment in much of the primary materials industry, little supplier information is being generated. While some man- ufacturers develop their own data bases, others find that largely empirical trials are the least expensive approach (from a local point of view). The scientific community has been reluctant to get involved in data-generating efforts that involve "no new science." Federal funding agencies, reinforced by peer review systems, also have shunned this area. Progress in generating this data could have a significant impact on a variety of industries and process applica- tions. Another recurring concern is the frequent lack of domestic suppliers for new materials, such as some ceramics. Manufacturers are reluctant to begin using new materials systems when only one
83 or perhaps no domestic supplier exists. Many new materials are initially imported in quantity from Japan or Europe. A related problem is the growing tendency for manufacturers and wholesalers to limit inventory. Indeed, limiting inventories has emerged as a smart manufacturing practice, especially with high interest rates. However, this practice grossly limits the availability of new and many old materials for manufacturing trials. In fact, most of the materials in reference handbooks are not available in tryout quantities. Lastly, there is an important interaction between the use of new materials and recycling and scrap practices. This is particu- larly the case in the automobile industry, in which a by-product of primarily steel construction has been the relative purity of car bodies as a source of steed scrap. The ease of recycling car bodies is being compromised by the materials substitutions now occur- ring, which could significantly increase material costs in a number of industries. Although these problems slow progress, none is sufficient to prevent increased use of new materials in manufacturing if those materials are cost effective in production, performance, and main- tenance. Limited availability is probably the greatest handicap, because the machining, tooling, and processing required for many new materials can be vastly different from those for traditional metal cutting; significant research in a production environment is a prerequisite for increased use. Fortunately, enough produc- tion experience is being accumulated with many new materials, particularly polymers, to demonstrate their advantages and to encourage efforts in other areas of materials research. Despite the handicaps, significant breakthroughs can be expected so that changes in manufacturing materials will keep pace with the many other developments on the factory floor. MATERIAL HANDLING TECHNOLOGY TRENDS This section will assess the major trends in material handling technology. Material handling systems are used to enhance hu- man capabilities in terms of speed of movement, weight lifted, reach distance, speed of though", sensory abilities of touch, sight,
84 smell, and hearing, and the ability to deal with harsh environ- ments. In this area, it is important to distinguish between equip- ment technology and design technology. Equipment technology is categorized by its primary functions: transporting, storing, and controlling materials. Transporting The material handling function of transporting material has been affected significantly by two trends toward smaller loads and toward asynchronous movement. The former is the result of the drive to lower inventory levels through just-in-time pro- duction. It has been manifested in the development of numerous equipment alternatives that have been downsized for transporting tote boxes and individual items rather than the traditional pallet Toads. The inverted power-and-free conveyor, powered by linear induction motors for precise positioning and automatic loading and unloading, is one example of the trend to develop transport equipment for small loads. Automatic guided vehicles (AGVs) for transporting individual tote boxes are also being developed, as are specially designed conveyors and monorails for tote box movement. The use of asynchronous movement in support of assembly has existed for many years. For example, asynchronous material handling systems were prevalent in automotive assembly before the paced assembly line was adopted at Ford in the early l900s, and the concept has been applied recently in some European au- tomotive assembly operations. In the early 1970s, Volvo began using AGVs to achieve asynchronous handling in support of job enlargement. Asynchronous material handling equipment is often used to allow a worker to control the pace of the process. The trend toward asynchronous movement appears to be par- tially motivated by the apparent success of Japanese electronics firms in using specially designed chain conveyors that place the control of the assembly process in the hands of the assembly op- erators. A workpiece is mounted on a platform or small pallet which is powered by two constantly moving chains. The platform is freed from the power chain when it reaches an operator's station.
85 After work is completed on the workpiece, the operator connects the pallet to the chain, and it moves to the next station; if the next station has not completed its work on the previous piece, the pallets accumulate on the chain. Asynchronous alternatives include using AGVs as assembly platforms and for general transport functions; "smart" monorails for transporting parts between work stations; transporter con- veyors to control and dispatch work to individual work stations; robots to perform machine loading, case packing, palletizing, as- sembly, and other material handling tasks; microload automated storage and retrieval machines for material transport, storage, and control functions; cart-on-track equipment to transport material between work stations; and manual carts for low-volume material handling activities. Storing The major trends in material storage technology are strongly influenced by the reduction in the amount of material to be stored and the use of distributed storage. Rather than installing eight to ten aisles of automated storage and retrieval equipment, firms are now considering one- and two-aisTe systems. Rather than being designed to store pallet loads of material, systems are designed to store tote boxes and individual parts. Also, rather than a centralized storage system, a decentralized approach is used to store materials at the point of use. Among the storage technology alternatives that have emerged are storage carouse! conveyors; both horizontal and vertical rota- tion designs are available. Furthermore, one particular carouse! allows each individual storage level to rotate independently, clock- wise or counterclockwise. A further enhancement of the carousel conveyor is automatic loading and unloading through the use of robots and special fixtures. A number of microload automated storage and retrieval sys- tems have been introduced in recent years. The equipment is used to store, move, and control individual tote boxes of material. Rather than performing pick-up and deposit operations at the end of the aisle, the microload machine typically performs such oper
86 ations along the aisle, since it is used to supply material to work stations along each side of the storage aisle. Of particular interest has been the introduction of storage equipment for production applications that previously was used for document storage in office environments. The trend toward lighter loads has resulted in a shift of technology from the "white- colIar" environment to the "blue-collar" environment. Despite the apparent need for automatic storage and retrieval of individual items, few equipment alternatives are currently avail- able, and those that are have not gained wide acceptance. This particular void in the technology spectrum has existed for a num- ber of years and does not appear to be of current interest to ma- terial handling equipment suppliers. Controlling The ability to provide real-time control of material has ele- vated material handling from a mundane "lift that barge, tote that bale" activity to a high-tech activity in many organizations. The control aspect includes both logic control and the physical control of material. In the area of logic control, the ability to track mate- rial and perform data input-output tasks rapidly and accurately has had a major impact on material handling. In physical con- trol, automatic controls have been added to a number of material handling equipment alternatives, allowing automatic transfer and assembly. Perhaps the fastest-growing control technology today in ma- terial handling is automatic identification. Likewise, the expecta- tion is that the greatest impact in the future will come from the ap- plication of artificial intelligence to transporting, storing, and con- trolling material. Among the alternative sensor technologies avail- able to support automatic identification are a wide range of bar code technologies, optical character recognition, magnetic code readers, radio frequency and surface acoustical wave transponders, machine vision, fiber optics, voice recognition, tactile sensors, and chemical sensors. The growth in the use of bar code technologies is due to three developments: bar code standardization, on-line printing of bar
87 codes, and standardized labels. The standardization of codes and labels came about through a concerted effort by the user com- munity. The Department of Defense led the way with its LOG- MARS study; that success was followed quickly by a concerted ef- fort by the automotive industry (the Automotive Industry Action Group). Other industries that have standardized codes and labels include the meat packing, health, and pharmaceutical industries. Others, such as the telecommunications industry, currently are involved in developing counterpart standards. Additional developments in the control of material handling equipment include off-wire guidance of AGVs. AGV technology is one of the most prominent areas of current material handling research and implementation.7 Functioning as a mobile robot, the AGV is being given enhanced sensor capability to allow it to func- tion in a path-independent mode. Through the use of artificial intelligence techniques, the AGV will be able to perform more than routine transport tasks without human intervention. Using sophisticated diagnostics, it will be able to execute advanced tasks such as automatic loading and unloading of delivery trucks. A number of European and Japanese firms are making ma- jor investments in the development of future-generation AGVs. Ranging from vehicles capable of transporting loads in excess of 200,000 pounds to those designed to transport individual printed circuit boards, a number of new entries into the U.S. market are expected within two to three years. A related control development that will have a major impact on material handling equipment technology is interdevice commu- nications. The manufacturing automation protocol (MAP) being developed by a number of firms led by General Motors (discussed in the section Factory Communications and Systems Technologies, under the subsection Networks) is expected to provide the com- mon data transmission link by which many different types of man- ufacturing hardware, including material handling equipment, will communicate. The driving force behind this standardization effort is the desire for truly integrated manufacturing systems across the entire hierarchy of manufacturing.
88 Design Technology In addition to the development of new and improved material handling equipment technology, new thinking has emerged on the design of material handling systems. Specifically, computer-based analysis, including the use of simulation and color graphics-based animation, is being used increasingly to design integrated material handling systems. Interactive optimization and heuristics also are being applied in the design of material handling systems. Considerable research has been performed in developing performance models of a vari- ety of equipment technologies. Trade-offs between throughput and storage capacity, optimum sequencing of storages and retrievals, and the automatic routing of a vehicle in performing a series of order-picking tasks are some of the issues that have been addressed in an attempt to gain increased understanding of material han- dling in the future factory environment. DEVELOPMENTS IN MATERIAL TRANSFORMATION TECHNOLOGIES This section describes the technologies of individual com- puter-controlled equipment, from numerically controlled (NC) ma- chine tools and smart robots to computer-aided design and arti- ficial intelligence, developments whose impetus comes from rapid advances in microelectronics and computer science. Rapid devel- opments in very large scale integration of integrated circuits have reduced the size, cost, and support requirements of information and machine intelligence while greatly increasing its capabilities. Microelectronic technology in the future will be embedded in each machine toot and robot and at every node and juncture of com- puter and communication networks. The capabilities provided by this embedded intelligence will revolutionize operations on the factory floor. Machine Tools Although numerical control was invented and applied some 30 years ago, it continues to change the structure of machine tools in
89 ways that still are not fully appreciated. Computerized numerical control (CNC) has replaced the punched paper tape of the original NC tools. As machines were developed specifically for NC, the traditional lines separating machine types began to blur, and two new classes of machines began to develop. The first class, called machining centers, generally operates with a stationary workpiece and a rotating tool. Feed of the toot in relation to the work can be handled by additional movement of the tool, the work, or both. The last method is necessary for contoured surfaces and in complex cases may require more than three axes of movement, often five, and perhaps as many as eight. These machines primarily perform drilling, milling, and boring operations, but they also can tap, thread, and, when necessary, mill a surface that simulates work produced by turning. The second new class of machines developed as a result of NC has rotating work and a stationary too! (except for feed). The machines are called turning centers and resemble lathes. They primarily do internal and external turning, drilling (of holes on the center axis), and threading, but many are equipped with powered stations that permit off-center drilling, tapping, and milling. Both of these new classes of machines can be equipped with automatic tool-changing devices and often have automatic work loading, sensors to check on operating conditions, measuring de- vices, and other features that enable them to operate for long pe- riods on different workpieces with little or no operator attention. With such versatility, a machining center and a turning center working together can perform all of the basic cutting operations on virtually any part that falls within the operating size limits of the machines. A number of special cutting and finishing processes supple- ment the basic processes performed by these machines. These include gear cutting, shearing, punching, thermal cutting, grind- ing, honing, and lapping. Although NC was not generally applied to these operations as quickly as it was to the basic cutting oper- ations, it is now applied to machines for each of them. (Because shearing and punching are done on presses and usually on sheet or plate material rather than on the heavier workplaces used for
go cutting, they are usually cIassifiec! as metal-forming operations. Thermal cutting also falls in this class.) In addition to the application of NC to traditional metal- cutting operations, several new cutting technologies have become important in many applications. The most widely used of these is electrical discharge machining (EDM), in which the workpiece is precisely eroded or cut by electric pulses jumping between an electrode and the workplace in the presence of a dielectric fluid. Electrodes, usually made of brass or carbon, are machined to the desired form. Although the cutting process is slow, the machines operate with little or no attention, and EDM is an efficient method of cutting many types of dies. A major recent development in EDM is the wire cut machine, in which the electrode is replaced by a fine wire sprayed with dielectric fluid. The wire slices through the workplace as if through cheese, making shaped cuts as the workpiece table moves by NC. The wire is constantly moving between two spools so that, in effect, fresh electrode is always being used. Low-power lasers began to be used for precision measurement about 20 years ago. Higher-power lasers are now used for welding and for sheet and plate metal cutting. Within the past two years, precision machine tools that use the laser as a cutting too] have been introduced in the United States and Japan, both for drilling and for cutting contoured surfaces. Other new technologies in- clude the use of electron beams for drilling and welding and the use of a plasma flame for cutting. Parts can be formed from sheet or plate in a variety of presses that bend,fold, draw, punch, and trim. The average age of presses currently in use is much higher than the average age of cutting machines, and users have generally been slower to innovate, but some press-working shops have taken advantage of new technolo- gies. For example, some shops have installed lines in which coiled sheet is unrollecl, flattened, trimmed, and shaped into parts by stamping, bending, and drawing in a continuous series of opera- tions. Others have installed transfer presses which make finished parts from strip in a continuous series of operations. Much of the progress in forming has come through better control of the material to be formed.
91 The only extensive use of NC in presses has been in punch presses that combine tool-changing ability with two-axis position- ing of the work for punching, nibbling, trimming, anc] cutting with lasers or plasma flame. However, NC controls are now beginning to appear on some other types of presses. Tooling Cutting tools are macle from a variety of materials: high- speed steel; carbides of tungsten, titanium, and boron; oxicles of aluminum and silicon (ceramics); cubic boron nitride; en c] syn- thetic and natural diamonds. Major advances in cutting-too] ma- terials sometimes cannot be fully utilizer] until machines clesignec] to take advantage of their properties are generally available. Great progress in cutting tools has been macle by applying a coating of one material (in some cases, two or three coatings of different materials) onto a base material. The proliferation of toot materials and coatings has become so complicated that computer software has been developec] to aic] the process. The resulting tools last longer, stay sharper, and can be user] to cut hare] materials such as heat-treated steel anc] abrasive materials such as fiberglass. As combinations of materials and coatings produce a growing list of tooling options, the variety anc] volume of tooling require- ments can be expected to proliferate. New product clesigns anc] performance requirements, product anc] process specifications, and changing lot sizes will create an ever-increasing need to match specific tooling with specific production applications. To achieve the high-quality, close-tolerance production demanded in the mar- ketplace, manufacturers will require a large inventory of tooling to ensure that the optimal tooling is available for all production re- quirements. Combinec] with the increased expense of tooling macle with rare materials anc] precision coatings, the costs of meeting tooling requirements will become major factors in capital buciget · . · ~ng c recisions. Improvements also have been made in clie en c] moIc] materials. More important, however, is the change taking place in the way cries anc] molds are produced. Traclitionally, they have been macle
92 by experienced craftsmen with a great deal of time-consuming cut and try in the finishing stages. The combination of newer EDM machines and computerized programming of die-sinking machines is removing much of the cut and try from this process. Jigs, which serve to position the too} more accurately in rela- tion to the work for drilling or boring, can usually be eliminated when NC machines are used. In fact, one of the major early ad- vantages of NC was the elimination of the production and storage of jigs. Of course, if the jig also serves as a fixture to hold the work on the machine, that function is not eliminated on an NC machine. Fixturing Fixtures hold and locate the part being worked during ma- chining and assembly operations. The main considerations in fix- ture design are positioning the part in the fixture, securing the part while the machining operation takes place, positioning the fixture relative to the machine tool, positioning the cutting too! relative to the part, and minimizing set-up times. New fixturing techniques add flexibility and programmability to minimize set-up time, maximize the flexibility of the machine, and reduce storage requirements for fixtures. The characteristics of the fixture depend on the process being performed, the shape of the part, and the tolerances required. For example, the workpiece may be subjected to strong vibrations or torque forces during some operations such as milling, while the forces in assembly operations will be much smaller. The fixtures required for these two operations are quite different and virtually incompatible. When a variety of tasks are performed, a large number of fixtures must be developed, stored, and accessed a very expensive undertaking. The need for a large number of fixtures remains a problem even for flexible manufacturing systems (FMSs) that can quickly and efficiently machine a number of different parts within the same part family. The FMS can help reduce economic lot sizes and reduce the expense of keeping parts in inventory. Unfortunately, this advantage is restricted by the need to have different fixtures
93 for different parts. The cost of multiple fixtures can account for 1~20 percent of the total cost of the system, and the fixtures can sometimes cost more than the rest of the system. Clearly, the full advantages of an EMS cannot be realized without the development of flexible fixturing that can conform to different part types and machining operations. A number of major research efforts are focused on the problem of flexible fixturing, ~.nd several solutions have been proposed. One approach would be to automate the current fixturing process, which uses blocks and clamps to align parts accurately. Instead of skilled toolmakers, robots could be used to assemble fixtures on coordinate measuring machines (CMMs). The fixtures would be mounted on standard pallets, permitting robots to load and unload parts easily and allowing easy alignment with machine tools. The CMM could cost $200,000 and vision-equipped robots at least $100,000; the hardware for the fixtures themselves and the software needed to control the robots would add to these amounts. Although the present cost may be prohibitive, this approach would ensure accurate location of the workpiece and it could be used for both machining ~.nd assembly operations. Another approach partially encapsulates the workpiece in a Tow-meTting-point alloy prior to machining. Encapsulation has been developed specifically for milling gas turbine and compres- sor blades of irregular shape. The unmachined blade is precisely positioned in the encapsulation machine. Rapidly injected molten alloy surrounds the blade and provides the clamping face, pro- tecting the blade itself. After machining, the alloy capsule is mechanically cracked open. The problem with this approach is that the blade must be positioned accurately in the very expen- sive encapsulation machine, which requires a different die for each workpiece. This limits flexibility and adds expense. A third approach is programmable conformable clamps. De- veloped at Carnegie-Mellon University for machining turbine blades, the clamps consist of octagonal frames hinged to accept a blade. The lower half of the clamp uses plungers, activated by air pressure, that conform to the contours of the blade. A high- strength belt is folded over the top of the blade, pressing it against the plungers, which are mechanically locked in place. Accurate
94 alignment can be done manually or automatically with sensors. Although the clamps are limited in the types and sizes of parts they can hold and their large number of moving parts may reduce reliability, they are automatic and ensure accurate alignment. Another approach is the fluidized-bed vise, in which small spheres are held in a container with a porous floor through which a controlled air stream passes. The spheres behave like a fluid, conforming to the contours of even irregularly shaped parts; when the air flow is stopped, the spheres come together to form a solid mass that secures the part. The advantages of this approach are that a variety of part shapes can be clamped, the clamping pro- cess is automatic, and the vise is inexpensive to build and operate. However, additional research is needed to establish a predictive mode! for the device and to eliminate the need for an auxiliary device to determine the location and orientation of the workpiece in the vise. Research is also under way in which electrically ac- tive or thermally active polymers are used in an authentic phase change bed instead of the pseudo phase change of the air-sphere approach. None of these approaches offers the flexibility needed in terms of variety of applications, the types and sizes of parts that can be held, and expense. They also do not address the problem of lo- cating the workpiece. The first three approaches use mechanical stops or surfaces, and the fourth requires an additional measur- ing system; this problem may be overcome by combining flexible fixturing devices with sophisticated robots. Sensors As machine too! automation advances, the instrumentation on the machine becomes increasingly important. Most of the early problems with automation tended to be instrumentation problems. Sensors to determine what is happening and monitor- ing systems to evaluate the sensor information are both needed. The role of sensors in a manufacturing environment is to gather data for adaptive control systems-for example, to supply guid- ance information to robots or to provide measurements for qual- ity assurance and inspection systems. Sensors can provide auto
95 mated equipment with vision, touch, and other senses, enabling the equipment to explore and analyze its surroundings and, there- fore, behave more intelligently. Sensors are currently used in factories to provide different types of data, such as the bipolar on-off of a limit switch, the simple numeric data of a temperature sensor, and the complex data provided by a vision sensor. Vision sensors, for example, can be used to determine part identification, orientation, and mea- surement data. Other sensors, such as tactile, acoustic, and laser range-finding sensors, are being used to measure force and shape, provide range data, and analyze the quality of welding processes. Sensor technology is a very active field of research. Sensor research that shows promise for manufacturing includes m~crome- chanics, three-dimensional vision for depth sensing, artificial skin for heat and touch sensing, and a variety of special-purpose sens- ing devices.9 Some of the special-purpose devices have no human analog. Examples are the water vapor sensors being developed for use in sophisticated adaptive-controT algorithms and the opti- cal laser spectrometry probes that monitor chemical processes in real time. The use of adaptive closed-Ioop control systems in man- ufacturing has increased the demand for a wide variety of special- purpose sensors and has stimulated the demand for sophistication in general-purpose sensors such as vision sensors. Other research is focused on the analysis, interpretation, and use of the data provided by sensors. Through the use of VEST techniques in IC fabrication, intelligent sensors equipped with mi- crochips can process data even before it leaves the sensor. For example, research is under way on vision systems that can inspect IC wafer reticles. Research on this vision system is focused on the mechanical accuracy of positioning devices, on the interface to the CAD data base describing the reticle, and on modeling the fabrication process to predict what the vision system will see. The visual information itself must be interpreted to determine whether to accept the wafer under inspection or to identify the flaw and provide feedback to correct for any imbalance. This type of intel- ligent sensor will eventually be integrated into many elements of manufacturing. Model-based sensor systems such as these which use process,
96 CAD, simulation, and control algorithms are expected to provide manufacturing sensor systems of the future with very complex analysis capabilities. These analysis capabilities will far surpass the monitoring and control capabilities of human operators by being more sensitive, more precise in analysis, more rapid in feed- back response, and more precise in corrective action. They will allow the factory of the future to work to very fine tolerances while maintaining consistently high quality control, approaching zero defects. Smart Robots One of the most common uses of advanced sensors is to make robots smarter. The senses of a robot are the sensors in its work cell that provide information to the robot's central controller. The "intelligence" of the robot is determined by the combined capa- bilities of its controller, its sensors, and its software. Most of the robots in the worId's factories today have primitive controllers and software and few, if any, sensors. They mindlessly weld, paint, and pick-and-place, and some would continue to do so even if no ob- ject were present to paint, weld, or grasp. Such robots are locked into a predetermined program that does not adapt to unexpected changes in the work cell. In contrast, advanced robot systems have sensors that inform the robot of the state of its world, controllers that can interface with the advanced sensors, and software that can adapt the robot's program to reflect the changing state of its world. This is an example of adaptive behavior using a closed- loop feedback system; to a degree, it is what people do when they engage in behavior that uses the senses. It is expected that 60 per- cent of all robots, especially those used for inspection, assembly, and welding, will utilize vision, tactile, and other sensors within the next 10 years.~° Smart robots have many advantages. About one-third of the cost of a robot work cell is the fixturing that holds or feeds each part in precisely the same way each time. This cost can be saved by smart robots that can find the part they need even if it is askew, upside-down, or in a bin with other parts; it is easier to change a robot program than to change the fixturing. Smart robots will be
97 much more adaptable to product changes because they will have less fixturing to change. Smart robots will be even more adaptable to difl3erent tasks when they can easily change their end effecter for a drill, deburrer, laser, or whatever too! is required. State-of-the-art robot systems embody elements of adaptive control and are now coming into use in factories around the world. One example is arc welding robots whose welding path is planned with the aid of a vision system that determines the location and the width of the gap to be welded. The robot software then adjusts the path and speed of the welding tool as the welding progresses. Although the welding example shows how adaptive control enables a robot to perform a task with built-in variance, the variance found in arc welding can be foreseen easily and taken into account by a human engineer or programmer. Adaptive control for robots with less-structured tasks is still in the research stage. Robots are programmed through a special-purpose computer language. State-of-the-art languages allow the robot to perform limited decision making on its own from information obtained with its sensors. However, these programming languages are limited because they can neither interpret complex sensory data, as from a vision or tactile sensor, nor access CAD data bases to get the information they may need to identify the parts that they sense. Present languages are also robot dependent; that is, they do not allow the transfer of programs from one robot to another. This means that robots must be programmed individually by valuable highly trained programmers. New robot programming languages that address some of these limitations are in development in academic and commercial re- search laboratories. The new task level languages will allow robot programming at higher levels: the robot can be told what to ac- complish or what to do with the part, and it will determine the best way to accomplish the task. The benefits expected when the new languages reach the factory floor include reducing the cost of programming, facilitating the coordination of two or more robots working cooperatively, and enabling advanced sensors to interface with the new systems.
98 Computer-Aided Design Computer-aided design is not a new technology; it has already achieved wide acceptance and use in manufacturing design, and it has replaced traditional drafting techniques in other areas such as architecture and cartography. It is important to understand CAD as a technology because it interrelates with many of the other technologies described here. For example, CAD-type systems are now being used to program robots and NC machining centers. (Detailed descriptions of the interrelated roles of CAD and the CAD data base in the factory of the future are included in the sections to which they apply.) A CAD system is composed of a graphics terminal on which can be displayed a picture of the part being designed. Designers enter the part data by drawing on a graphics tablet connected to the computer. A keyboard is used to enter dimensions and other data. The part description is then stored as one of many such part descriptions in a CAD data base. The computerized part description is not a picture, but rather a representation of coordinate points and geometric shapes from which a picture can be constructed. A particularly successful standard, the Initial Graphics Exchange Standard, has been developed for transferring data from previously incompatible representations on one CAD system to another CAD system. (This standard will be described in the Data Bases section.) CAD offers many immediate benefits: parts can be rotated, scaled, and combined onscreen in three dimensions to enable de- signers to better visualize them; repetitive sections can be redrawn automatically; overlays can be easily shown onscreen; and engi- neering drawings can be easily updated and printed. Other, more significant, benefits over the long run concern the use of the data in the CAD data base. These data a com- puterized representation of the parts can be used by the engi- neering and process planning functions, saving much reentering of data, eliminating sources of human error, and opening up a great avenue for cooperative design that includes feedback from engineering and manufacturing. Also, if the CAD data base is the only and therefore up-to-date source of part specifications, it
99 eliminates a major current problem, concurrent use of multiple versions of part specifications. The microelectronics industry probably has the most inte- grated uses of CAD. A new microchip can be designed on a CAD terminal. Once the design is in the CAD data base, the chip's performance can be simulated, the design can be modified if nec- essary, and the masks for the chip can be made, all automatically from the data entered at the CAD terminal. Although other in- dustries have not yet achieved this level of integration, it has be- come an embodiment of the computerized design-test-modify-test- fabricate model that may change the way manufacturing entities are organized. Graphic Simulation As manufacturing systems come to include advanced systems such as smart robots, it becomes more and more important to be able to simulate their behavior. Two distinct kinds of simulation are now being used in manufacturing. One is the simulation, often graphic, of a single process, robot, or work cell. The second is the simulation, generally mathematical, of a system such as an FMS, a new or modified production line, or an entire factory. The former may be regarded as important in tactical or local decisions, the latter in strategic or system decisions. For this reason, graphic simulation will be treated here, and mathematical modeling will be covered later in the communications and systems section. Graphic robot simulation is beginning to be used to select the most appropriate robot for a particular task or work cell and to plan the cell layout. The production engineer can use simula- tion to reject robots that do not visually appear to suit the task because of their arm configuration or timing constraints. Graphic simulation is also used for visual collision detection in the work cell, but this method is prone to error and not recommended. Some vendors of graphic robot simulators have adapted their software to generate actual robot control programs, which is termed offline programming. It permits the development of robot programs without shutting down a productive work cell, thus al- lowing efficient, concurrent work cell design. Although programs
100 developed off-line currently must be used on the specific robot for which the system was designed, research to include a variety of robots in the simulation system is being conducted. For example, an ofl~-line robot programming system has been developed that can simulate any of six commonly used robots. Researchers are also working on the related problems of how to simulate a com- plex sensor, such as a vision sensor, and how to debug an off-line robot program that makes decisions based on advanced sensory input. Graphic simulation and off-line programming prorn~se to provide cost, time, and personnel savings in the efficient design of programs, work ceils, and processes. Artificial Intelligence Artificial intelligence is a set of advanced computer soft- ware applicable to classes of nondeterministic problems such as natural language understanding, image under- standing, expert systems, knowledge acquisition and rep- resentation, heuristic search, deductive reasoning, and planning. ~ ~ Artificial intelligence (AI) technology will emerge as an inte- gral part of nearly every area of manufacturing automation and decision making. Research that will affect manufacturing is be- ing conducted in several areas of Al, including robotics, pattern recognition, deduction and problem solving, speech recognition and output, and semantic information processing. As with sim- ulation, AT will be used at different levels in the factory of the future. Most of the Al applications will be integrated into the software that controls automated machinery, record keeping, and decision making. Artificial intelligence is not a new field, but the maturing fruits of 20 years of Al research are just now becoming available for commercial applications. The types of Al products that will have a significant impact on manufacturing include · expert systems in which the decision rules of human ex- perts are captured and made available for automated decision making;
101 · planning, testing, and diagnostic systems; and · ambiguity resolvers, which attempt to interpret complex, incomplete, or conflicting data. The AT applications that deal with individual machines, pro- cesses, or work ceils are described here; those that deal with system-level decision making will be integrated into the Factory Communications and Systems Technologies section. Expert systems are in productive use today in isolated indus- tries; petrochemical companies, for example, use expert systems for the analysis of drilling samples. Digital Equipment Corpo- ration has used an expert system for a number of years, saving several million dollars annually in configuring the company's VAX computer systems. As human experts with years of experience be- come scarce, the expert system provides a way in which to capture and "clone" the human expert. An interesting feature of expert systems is that they can explain the train of reasoning that led them to each conclusion. In this way, the systems also can be used to augment human decision making, in much the same way as med- ical expert systems have been used. Current expert systems are best suited to situations that are somewhat deterministic when the expert's rules are known. For this reason, rapid emergence of expert systems can be expected in limited areas of technical knowledge such as chip design, arc welding, painting, machining, and surface finishing. In the l990s, expert systems are expected that will learn from experience; this means that expert systems eventually will be developed for specialties in which there are no human experts. Although still primarily in the laboratory, one type of AT software is attempting to simplify the use and expand the appli- cability of programmable equipment. For example, advanced user interfaces are now being developed that use "natural language," so that a manager can type a request at his work station in more-or- less plain English. The Al software will determine what he means, even if the request has been phrased conversationally or colloqui- ally, and provide interactive assistance for decision making. By the year 2000, managers will probably be communicating with their work stations by voice, another application of Al techniques. Ar
102 tificial intelligence technology promises to make it much easier for computers and computerized equipment to be used by personnel not having computer training, such as managers, engineers, and operators on the factory floor. FACTORY COMMUNICATIONS AND SYSTEMS TECHNOLOGIES In contrast to the materials and process technologies des- cribed above, communications and systems technologies tend to operate at higher levels, allowing previously separate areas of man- ufacturing to be integrated into systems of manufacturing. A man- ufacturing system is defined as a system created by the intercon- nection and integration of processes of manufacturing with other processes or systems. This definition implies that manufacturing systems vary from a basic system, which couples a few processes, to a hierarchical system, which integrates lower-level manufactur- ing subsystems into the single aggregated system. Such a system is termed a computer-integrated manufacturing (CIM) system (Fig- ure Am. This variation in complexity and level makes the concept of a manufacturing system elusive to grasp. It may be helpful to think of it as an approach, a systems approach, to incrementally integrating the functions of the manufacturing corporation. The major characteristic of manufacturing systems is their sharing of information, their communication. Traditionally, man- ufacturing information has been created and communicated by humans writing on paper. This paper information was based on the understanding of the human expert at that moment, al- though often that understanding did not accurately reflect the real state of the factory at that moment. This paper method is people intensive, time-consuming to create and distribute, often inaccurate, and in frequent need of revision. As an information communication method, it virtually guarantees delay, inaccuracy, and expense. The advent of computer technology and network communi- cations is changing the face of the factory floor, much as office automation has changed the front office. This technology permits
103 .° c, r--~ add ._ ~ 1 1 1 c 1 L=l r O ~ . ~ ~ c 3 Q ~ ._ O UJ , ~ t tar . ~ 4- Q ~ C' Z O ~ Cot a, ~ O a) O Cut ~ _o ~0 in ~ ._ ._ ~ lo: 1 ~ ' 1 T . I L__] also'= ~1 m ~1 1 1 1 4 - , ~ ~ . O cot en a) 4= i OD ~0 . A v w w 1 ~ 1 ~ w Ma . i- o v w . o . 4. v O o . 4= w L' ~0 ,
104 the system to generate its own data according to the information provided by real-time sensors built into automated machining, as- sembly, and inspection stations. The system gives the data to a computer, which interprets the data and takes appropriate action. This action may be to control the machining process, to replace a worn tool, or to decide whether to communicate the data, to whom, and how much. This automated creation and sharing of information avoids the present duplication of data in several files or data bases, and it collects and communicates data at a scale and speed that will create opportunities in manufacturing never before available. For example, computer-controlled feedback permits a system to be self-diagnosing, self-maintaining, and eventually self-repairing. It allows the collection of statistical data that can be used for imme- diate adaptive feedback, quality control analysis, and the produc- tion of trend data. More important than any single benefit, this sharing of information makes possible the linking of systems into system aggregates. Previously disparate systems may be linked horizontally, and hierarchical adaptive control and reporting sys- tems may be created by integrating vertically. This information, or data, integration is the synergistic key to building manufacturing systems with broader scopes and at higher levels. The long-range goal of the manufacturing systems approach is CIM the complete integration of the manufacturing subsystems that operate on the factory floor, the tie-in of tech- niques of optimization, mathematical modeling, scheduling, and data communication with the other functions (accounting, mar- keting, etc.) in the total manufacturing enterprise. Note that manufacturing systems are at once a means and an end. Systems of manufacturing are integrated through the appli- cation of several technologies: communication networks, interface development, data integration, hierarchical and adaptive closed- loop control, group technology and structured analysis and de- sign systems, factory management and control systems, model- ing and optimization techniques, and flexible manufacturing sys- tems. Artificial intelligence techniques will be embedded in, and inseparable from, most of these technologies. Communications technologies those associated with networks, interfaces, and data
105 bases-may be the most critical to U.S. manufacturing progress because they are the keys to the immediate development of manu- facturing systems. On the other hand, technologies that analyze, manage, and optimize the system hold the greatest promise for improving the long-term competitiveness of U.S. manufacturing. These technologies will facilitate progress towards the goal of to- tal integration from design to delivery. Each will be described in depth, both individually and as they relate to the full CIM concept. Networks The manufacturing network will be the backbone of factory communications and, therefore, of factory automation. Communi- cations between tightly coupled components, such as robots and sensors, and between elements of an FMS require that data be exchanged in real time. As the complexity of the factory sys- tem increases, including the linkage from design to planning and production, the need for factory communications will continue to expand. Networks provide the physical mechanism for this com- munication between heterogeneous systems. The network must not only transmit the raw data but also retain its meaning, so that a different computer, running a different program, may use it. The goal of CIM is to allow Al manufacturing activities us- ing heterogeneous hardware to communicate as though they had a common language. Networks currently provide a protocol, or agreed-upon stan- dard, for computer communication. Most major computer ven- dors, as well as manufacturing equipment vendors, have defined proprietary network protocols. Thus, a variety of incompatible networks, such as Ethernet and Modway, are in use in factories today. In addition to resolving the compatibility problem, ad- vances in network architecture are required to meet the specific communication needs of manufacturing. The speed and traffic re- quirements of manufacturing communication must be taken into account, as well as provisions for interfaces between otherwise in- compatible networks. General Motors and its major vendors began work on a set
106 of manufacturing automation protocols (MAP) for this purpose several years ago. The development of MAP has been broadened recently to include support by more than 100 major manufactur- ers and universities, including the National Bureau of Standards (NBS). These otherwise competitive groups realize that no single vendor can meet all the needs of a manufacturing system and that MAP may provide a solution to the communications problem be- tween their equipment and other vendors' machines and networks. MAP is an attempt to define the seven-level communications protocol proposed by the International Standards Organization. Although all seven levels have not yet been standardized, vendors are already selling MAP-compatible products, and farsighted pur- chasers are demanding that their new hardware be MAP compat- ible. A recent breakthrough by Industrial Networking, Inc., has put MAP on a single microchip, which will facilitate the devel- opment of factory communication networks among heterogeneous machines. While significant challenges remain, the broad member- ship and participation in the MAP effort can be used as a model for specifying and solving other manufacturing system problems. Interface Standards The network is expected to provide the physical and log~- cal path for data communication in a factory system, but much more is required for effective communication. Networks provide the physical language and format, but do not address the seman- tics or effective use of the information communicated. Interface standards are needed to facilitate the effective communication of meaningful data. The key to data integration is standardization that does not stifle innovation. Standardization of data representation within the data base is necessary to allow the full meaning of the data to be retained even when it is communicated. Current practice requires vendors of systems or modules to provide special-purpose interface definitions for each pair or family of modules that com- municate. However, in some areas standards have evolved through the cooperation of users and vendors. Examples are the CEDATA file for NC machines and the Initial Graphics Exchange Standard
107 (IGES) for CAD data base information exchange. IGES has en- abled previously incompatible CAD systems, with data stored in radically different formats, to communicate that data while pre- serving most of the meaning. Yet these standards rapidly grow out of date as technology moves forward. CEDATA is inadequate for nondeterministic (sensor-basect) machine too! programs, and IGES does not work on solid-modeling CAD systems. The IGES continues to evolve, pointing the way to wider data integration. The challenge is to define standards that will withstand the de- mands of continued factory innovation or to establish mechanisms to update standards as needed. Standards are also the solution to the interface compatibility problem that arises when equipment from different vendors is used in a network. The interface connects one machine to a communi- cations system, which is connected to other machines, computers, and communications systems. The RS-232 interface standard is a simple protocol that has allowed communication between het- erogeneous microcomputers and between computers and a host of other devices. Many machines already come with the limited RS-232 interface, but more progress is needed in standardizing manufacturing interfaces. MAP includes the definition of intelli- gent interfaces which can connect previously incompatible systems to a network. The lack of interface standards can be a major impediment to achieving CIM. If well-defined information interfaces between modules or subsystems were established for the components of manufacturing systems, components could be developed indepen- dently and enlarged as advances in technology became available. This would facilitate compatibility of the equipment of multiple vendors in the heterogeneous systems expected in the factory of the future. Interface standards of this type are the basis of re- search at the Automated Manufacturing Research Facility at NBS. Significant questions must be answered, however, before the infor- mation interface for the modules in the manufacturing system of tomorrow can be defined. In addition to these interface standards for information, two other kinds of interface standards are needed. The first and most neglected is the interface between human and programmable sys
108 tems. The second is the physical interface between mechanical systems. The man-machine interface includes the commands to be giv- en by the human to make the machine perform a task successfully and the input device or physical method keyboard, joystick, light pen, or voice- for entering those commands. Most current pro- grammable systems are commanded through a programming lan- guage that is proprietary to the vendor of the system. This has given rise to a Tower of Babe! of control languages requiring highly trained programmers to control modern manufacturing systems. No programmer can begin to master all of the languages and input devices found in an automated factory. Two recent trends are expected to ease the interface prob- lem between nonprogrammers, such as engineers and technicians, and the increasingly complex programmable automation systems found in the areas of robotics, NC tools, material handling, and processing systems. Hierarchies of languages and personnel are be- ing developed in which highly trained programmers will deal with the raw control languages and sophisticated control algorithms, less-skilled programmers will deal with a higher-level simplified language, and equipment operators will not use actual program- ming languages at all. This hierarchy automation systems pro- grammer, applications programmer, user programmer, and user- parallels the evolution of personnel in the computer field. The second trenc] is the use of Al to develop task-level control languages (discussed above in the Smart Robots section). Cur- rently under research, task programming systems will reduce pro- gramming requirements to the steps on a common process plan- ning sheet so that programmable manufacturing systems of the future will be controlled by statements similar to those one would give to a person doing the same task. These advances will pro- vide new generations of specialized, user-friendly manufacturing subsystems that will make the most of factory personnel. The mechanical interface problem for the factory is solved most easily by the development and adoption of standards. The lack of standards for the newer systems is a major impediment to progress. Examples of mechanical interfaces in need of standard- ization include
109 · robot end-o£arm and gripper attachments; · pallets, totes, and other part conveyances; · the mechanical interface for the loading and unloading of parts and pallets at machining centers; and · the interface between robot carts and material handling systems. Progress with the mechanical interface problem requires the usual consortium to agree on and promulgate standards. The major roadblock has been the lack of an organized body, leadership, and focus on the problem. Data Bases Network and information interface standards are the means of sharing data, but it is not enough to move data from one appli- cation to another. The data must be stored, and the semantics, or full meaning, of the data must remain intact when the data are retrieved. A data base provides the long-term memory, or storage facility, that contains the manufacturing data, and the informa- tion retrieval system extracts specific data from the data base. Current practice finds a large number of information retrieval systems in place even at the same company. Each system is as- sociated with one major function, such as accounting, shop floor information, material requirements planning, or quality control statistics. Many of these data systems are state-of-the-art infor- mation retrieval systems with complex functions to maintain and update the data base. Unfortunately, the different data bases contain redundant and conflicting data in incompatible formats. Furthermore, many of these systems are dedicated to a single com- puter and use proprietary data representations that are incompat- ible with those of other systems. Thus, the task of developing an integrated manufacturing data base management system that can include every major function in a factory is formidable. The most serious immediate barrier to the integration of man- ufacturing data is the incompatibility of CAD data with informa- tion needed by computer-aided engineering (CAE) and process planning. CAD technology has become highly effective in captur
110 ing the geometry of parts, including the description of dimensions, shapes, and surfaces. Real parts, however, are made up of smaller components and may themselves be components of a larger assem- bly. The CAD data base currently cannot capture the relationship of the parts to the whole, but both CAE and process planning require detailed attention to the joining of separate parts, their mating surfaces and tolerances, and their overall dimensions ad ter assembly. The CAD data base does not include knowledge or specification of materials, but CAE needs material data for its engineering analyses, process planning needs it in the creation of NC programs, and material handling needs it to select material from inventory. A second serious barrier to the integration of manufactur- ing data is the current inability to mode] the processing portion of the overall system (the two right-hand boxes of Figure A-~. Such a model would allow information on production costs and capabilities to be fed back, on-line, to the product design activity as it is being performed. This capability is essential to optimiz- ing the producibility of products at the design stage. With this capability, each decision proposed in the engineering design pro- cess would result in simultaneous information on the effect of that decision on production costs and required capabilities (relative to available capabilities) for production of the product. It also would result in major cost savings in the production activity, since it is well known that the majority of production costs are frozen at the engineering design stage. While such computer-based integration of manufacturing data is technologically feasible, many difficult problems must be solved to bring it into being. A further problem with the data in a CAD data base is that the geometrical data cannot be searched or aggregated in the ways that have become standard for textual data. Without explicit hand coding, it is not possible, for example, to retrieve all parts which use a particular fastener. Group technology (discussed later in this section) is an attempt to code and classify the geometry, function, and process data in a way that will permit the use of standard retrieval functions. One of the keys to the data integration problem lies in the de- velopment of flexible data schemes. A schema is a method of stor
111 ing data so that its meaning and accessibility are retained. Most data bases use rather fixed data schemes that restrict the new types of information that may be added and limit data retrieval capabilities. Future data bases will have more flexible schemes so that, for example, materials information can be added to the CAD data base by an engineer at a CAE station or by an expert system that contains knowledge of materials and applications. Beyond the compatibility problem are other technical chal- lenges to the implementation of manufacturing data base systems. For example, experts predict that future manufacturing data bases will be 20-50 times larger than present data bases. The size of the data base, the time tolerances for communication, and the vari- ety of users suggest that a manufacturing system data base will be distributed across multiple heterogeneous systems, which may be in different geographical locations. This presents significant technical challenges to the achievement of a logically integrated manufacturing data base. The concepts and protocols normally used to ensure proper access, control, and update will need to be expanded to meet this sophisticated method of data base organi- zation. Interim solutions in place today are neither geographically nor heterogeneously distributed, but progress toward these goals is being made. A last challenge posed by the manufacturing data base is the use of probabilistic or incomplete data. Current data base systems can only represent facts and cannot deal with uncertainty or conflict within their data. Manufacturing information systems of the future will depend on Al to deal intelligently with this type of information. One of the most critical roles of people in the factory of the future will be to interact with intelligent manufacturing systems through work stations, terminals, or networked microcomputers. As expert systems and other forms of Al become embedded in sys- tems of manufacturing, the systems will be able to perform more and more of the decision-making tasks previously performed by people. At first, these automated decisions often will have to be reviewed by people and then interactively modified, much as an architectural plan takes shape in a dialogue between client and architect. People without knowledge of the data schema will rou
112 finely query the system for information needed to make decisions. The data retrieval system will have to determine exactly what is important to the inquirer and then retrieve and massage the ap- propriate data. The person may even want the system's "opinion," or the system may ask for the person's opinion. The factory of the future will regard personnel and intelligent systems as partners in a dialogue that should encourage very sound decision making. Group Technology Group technology~3 (GT) is a key philosophy in the planning and development of integrated systems. In practice, GT is defined as a disciplined approach to identifying by their attributes things such as parts, processes, equipment, tools, people, and customer needs. These attributes are then analyzed to identify similarities between and among things; the things are grouped into families according to similarities; and these similarities are used to increase the efficiency and effectiveness of managing the manufacturing process. Although it is relatively simple to define GT, it is difficult to create and install a GT system because of the difficulty in defining clearly how similar one part is to another. For example, parts can be categorized in terms of shape or manufacturing process requirements. These two different viewpoints require a flexible approach to the GT data base and the realization that parochial, departmental views of coding may allow some localized cost saving but miss the large corporate savings possible. The GT concept requires that the attributes of a thing, such as a part, be identified and classified. Attributes can be visual, such as the surface finish or shape of a part; mechanical, such as the strength of the material; or functional, such as the clock aspect of a printed circuit board. The attributes may also be related to the environment of the part, such as the processes or equipment necessary to make it. Because of the many possible coding strate- gies, it is hard to know in advance exactly what attributes will be important as the GT data base is used by more and more kinds of software. It is therefore important to guarantee that the data
113 base structure is flexible enough to add attributes and to modify coding schemes as necessary for new applications. The four basic GT applications are design retrieval, FMSs, purchasing support, and service depot streamlining. Design re- trieval is a GT application in the design engineering area to pro- vide the maximum potential for part standardization. Moreover, it permits greater cooperation between the design engineer and manufacturing engineer by providing feedback about specific part attributes in the GT data base. Design retrieval will also sup- plement product reliability data based on the actual performance of parts with similar attributes. Finally, by determining the re- lationships of new parts to previously designed parts, it increases human productivity in the generation of new designs and the re- vision of old designs. The software necessary to implement design retrieval is the simplest of all GT software; it involves a simple query to a GT data base for specific feasible ranges of variables. Flexible manufacturing systems exemplify a more sophisti- cated and more profitable GT implementation. Such an FMS can be created by identifying a cluster of machinery that can, or will be able to, service a particular family of parts. The FMS can then be streamlined to produce this part family optimally. Purchasing support is a rather new GT application, yet al- most every major manufacturer has a quasi-GT system, called a commodity code, already performing this task. A rigorous GT system may pay a tremendous financial reward by permitting all related parts, including those that do not obviously belong to the same families, to be identified throughout a factory or corporation. In one instance, a vendor of hoses offered a 50 percent reduction on hose prices if a corporation could identify all hoses and their attributes to be purchased over a given period. The corporation saved millions of dollars by rigorously following a GT system to identify all hoses to be ordered. A GT purchasing support system offers buyers a significant way to cut costs through knowledge and buyer leverage. Service depot streamlining is a GT application which can help determine the most advantageous service parts strategy by identi- fying where alternate parts may be used. Standardization of parts
114 in the service depot allows for substantial reduction in inventory and repair time, even if the standard is the most expensive item. The premise of a GT system is that the similarity of parts and processes can be turned into substantial cost savings. The most far-reaching applications of GT will be made possible by the structuring of the parts data base itself. If the data base informa- tion is captured in attribute form and linked to applications by similarity, the ability of the data base to support manufacturing decision making will be greatly enhanced. This global application of GT to data base design is only now gaining popularity, and research on it is still in its infancy. One of these far-reaching applications is in the area of pro- cess planning. By performing a rigorous analysis of manufacturing processes and parts to be made, a manufacturer will improve his ability to move from the present method of process planning into the more highly integrated future. For example, a manufacturer may evolve from his present variant process planner, which uses GT to match part families to process families, to a more sophisti- cated generative system in which more knowledge captured in the GT data base will be used to optimize the process plan. Eventu- ally, a highly integrated system can be achieved that delays the final process planning step until the part is to be made, opti- mizing not only the process but also capacity utilization. Group technology, like network technology, will be a cornerstone of CIM systems. Adaptive Closed-Loop Control One difference between an automated system and an intelli- gent system is the amount and kind of feedback that is generated from an activity and passed up to a decision-making entity. This feedback allows a system to know its own state, to know when it is out of balance, and to respond to the imbalance until sta- bility is achieved. This adaptive closed-Ioop control will be used at all levels of manufacturing systems from sensor-based feedback to robots or NC tools, to inspection station-based feedback to a cell controller, to a factory floor data collection system that feeds back to process planning and scheduling systems. This property
115 of adaptive feedback is the key to improving product quality, with zero defects a realistic goal (see Figure A-1. Adaptive feedback is also one key to better management, since only in this way can a manager know exactly the state of production, including exact costs. Systems of manufacturing have a feedforward property that will allow management to control the factory floor with an effectiveness and immediacy never before pos- sible. (A more global discussion of hierarchical control is found in the subsection Computer-Integrated Manufacturing Systems.) Feedback and feedforward properties can also provide ma- chines and systems with sel£diagnosis features. Thus, a manu- facturing system can tell if something is wrong with it and what is wrong and can suggest the remedy to a higher entity. For ex- ample, researchers at the AMRF have already demonstrated the ability to sense when a too] is about to break so that automated equipment can change the too! without the disruption caused by untimely failure. Next will be limited self-maintenance and repair capabilities. When a data-intensive system breaks down, the in- tegrity of that data is threatened. Self-diagnosis will inform the data base system of the integrity of the data and, if it is threat- ened, the system will take either conservative or remedial action. Factory Management and Control The factory of the future will be managed and controlled through automated process planning, scheduling, modeling, and optimization systems. The successful implementation of large- scale factory-level systems depends upon structured analysis and design systems that depend heavily on GT. Limited structured analysis systems, such as the Air Force-sponsored Integrated Com- puter-Aided Manufacturing Definition, have been in use for years in the analysis and design of large projects. Only through such sys- tems can a manager know the exact state of his factory, and only through such exact knowledge of the present can a manager intel- ligently implement systems of manufacturing for the future. New systems development methodology packages, such as STRADIS, promise help in this area, but much work remains to be done be- fore such systems are easily used by the actual decision makers.
116 Similar work must be done on process planning and scheduling systems before they can use the feedback and feedforward prop- erties of the hierarchical and adaptive closed-Ioop control systems to be found in future manufacturing systems. Management functions will be hierarchically distributed so that "go" an;d "halt" decisions may be made effectively from many levels and by human or machine. Through the use of terminals on the factory floor and throughout the decision-making structure, the system can respond instantaneously to human command. At first, most of the decision making will rest in the hands of humans. Low-leve} manufacturing systems now work in this way. As the systems become integrated at higher and higher levels, decision rules and methods will be built into them; systems wiD develop plans to carry out human-specified activities. On the authorized human's approval, the system will carry out the task, making low- leve] decisions on its own. If a low-level decision-making entity does not have a certain level of confidence in its decision, it may pass the decision up to the next-higher entity, be it human or computer. This new kind of man-machine interaction will allow humans to do what they do best: create, define, and communicate. The machine will do what it does best: work hard, steadily, and accurately. Modeling and Optimization Systems One tried and true method of representing an activity to a computer is through mathematical modeling. Computerized mod- eling tools, such as SLAM, have been used by simulation experts for years, but simulation is still more an art than a science. With experience and feedback, our ability to represent complex activ- ities mathematically will be refined. A modeling and simulation package will be a necessary part of an intelligent structured anal- ysis and design system. It is hard to overstate the importance of structured analysis and design systems; they will operate at high levels, with much built-in decision making. System modeling will become a commonplace and necessary prerequisite to the successful design and implementation of large- scale manufacturing systems. This is because large projects con
117 tain too many facets to be managed effectively with only hu- man memory and computation capability. Artificial intelligence techniques will be needed to reduce the tremendous amounts of data generated by such systems to humanly understandable terms. This intelligence must be of a higher order than the Al expert sys- tems in existence today. With the addition of Al, a modeling system can become an optimization system, guiding its human managers to the most productive, most cost-effective, or highest-quality utilization of re- sources. With such optimization capability, managers will be able to sit at their work stations and, in real time, analyze the vari- ous possibilities to determine optimal solutions and mixes. The availability of accurate information on cost, time, and quality will eliminate much of the guesswork in manufacturing decision mak- ing. Flexible Manufacturing Systems Flexible manufacturing systems are expected to dominate the factory automation movement within 10 years. These FMSs will be tied into larger-scale manufacturing systems, but it is valuable to consider the FMS as a critical unit or building block in total factory integration. An FMS may be described as an integrated system of machines, equipment, and work and too} transport ap- paratus, using adaptive closed-Ioop control and a common com- puter architecture to manufacture parts randomly from a select family. The hardware components of an FMS may include an NC tool, a robot, or an inspection station. The part family processed by the FMS is defined by GT classification. For greatest produc- tivity, the FMS is optimized to produce only one family of parts, and conversely, the parts produced by the FMS are designed to facilitate processing by the FMS. The concept of flexibility as used in an FMS includes parts; · use of GT to achieve a part mix of related but different batching, adding, and deleting of parts during operation; dynamic routing of parts to machines;
118 and rapid response to design changes; making production volume sensitive to immediate demand; dynamic reallocation of production resources In case of breakdown or bottleneck. Flexible manufacturing has been a reality in U.S. industry since its introduction in 1972, and the number of new FMS instal- lations is doubling every two years. The number and flexibility of FMSs is expected to increase, and the cost of FMS installations is expected to drop. Although the United States was largely re- sponsible for the technological development of the FMS, Western Europe and Japan both have more FMS installations than this country. In fact, one of the most frequently cited FMS instal- lations is located in the Messerschmidt-Boelkow-Blohm (MBB) plant in Augsburg, West Germany. The basic elements of this FMS are 25 NC machining centers and multispindIe gantry and traveling-column machines; automated too! transport and tool- changing systems; an AGV workplace transfer system; and hier- archical computer control of all these elements. The FMS is used to build wing-carrythrough boxes for Tornado fighter-bombers. Comparisons by MBB of the performance of this FMS versus the projected performance of stand-alone NC machine tools doing the same work clearly show the advantages of the FMS approach: number of machine tools decreased 52.6 percent; workforce reduced 52.6 percent; · tooling costs reduced 30 percent; · throughput increased 25 percent; · capital investment 10 percent less than for stand-alone equipment; and · annual costs decreased 24 percent.~4 U.S. statistics for FMS installations are no less startling. An FMS at General Electric (GE) improves motor frame productivity 240 percent; an AVCO FMS enables 15 machines to do the work of 65; and at Mack Trucks, an FMS parrots 5 people to do what 20 did before. In addition to productivity enhancements, the FMS offers increased floor space capacity. GE, for example, reported
119 that floor space capacity was increased 50 percent, with a net floor space reduction of 30 percent. An FMS can make a factory more responsive to its market GE reported a shortening of its manufacturing cycle from 16 days to 16 hours. Computer-Integrated Manufacturing Systems The technologies discussed above will be integrated to create CIM systems whose synergy will make the whole greater than the sum of its separate technologies. CIM systems promise dramatic improvements in productivity, cost, quality, and cycle time. How- ever, since full CIM has not yet been accomplished and depends on continued technological progress, the benefits are difficult to quantify accurately. Incremental gains from the implementation of individual technologies and subsystems will be substantial. These benefits are illustrated by the following data from five companies that have implemented advanced manufacturing technologies over the past 1~20 years:~5 Reduction in engineering design cost Reduction in overall lead-time Increase in product quality Increase in capability of engineers Increase in productivity of production operations Increase in productivity of capital equipment Reduction in work-in-process Reduction in personnel costs 15-30 percent 30-60 percent 2-5 times 3-35 times 40-70 percent 2-3 times 30-60 percent 5-20 percent The cumulative gains of total system integration can be expected to build on these results exponentially. The long-range goal of CIM is the complete integration of all the elements of the manufacturing subsystems, starting with the conception and modeling of products and ending with ship- ment and servicing. It includes the tie-in with activities such as optimization, mathematical modeling, and scheduling. A CIM system is created by the interconnection or integra- tion of the processes of manufacturing with other processes or
120 systems. The resultant aggregate system provides one or more of the following functions or characteristics: · An information communication utility that accesses data from the constituent parts of the system and serves as an infor- mation communication and retrieval system. · An information-sharing utility that integrates data across system elements into a unified data base. · An analysis utility that provides a mathematical mode! of a real or hypothetical manufacturing system. Employing sim- ulation and, when possible, optimization, this utility is used to characterize the behavior of the modeled system in various con- figurations. A resource-sharing utility that employs mathematical or heuristic algorithms to plan and control the allocation of a set of resources to meet a demand profile. A higher-order entity that integrates information and pro- cessing functions into a more capable, effective processing system. These functions and characteristics are not mutually exclusive in actual manufacturing systems; rather, they overlap significantly with all of the elements interconnected and integrated continually to form a single aggregated CIM. Perhaps the most important and least understood step in this process is the creation of an integrated system that is a higher-order entity; this is the true system-building goal. Both horizontal and vertical growth of CIM systems can be expected as the year 2000 approaches. State-of-the art technol- ogy now includes small aggregates of computer-integrated tasks, often called islands of automation. Such islands of automation are found in design, where CAD work stations from different vendors share their data through a common data base and data conver- sion interfaces; in planning, with manufacturing resource planning (MRP) systems; and in production, where a work cell composed of a robot, machine tool, and inspection station may be coordinated by a cell controller. In leading-edge plants, several of these islands of automa- tion have been aggregated into larger manufacturing subsystems, termed continents of automation. At this level of integration, links
121 exist between the design and engineering departments, with CAD terminals and data bases sharing data with CAE work stations and data bases. In planning, MRP can be linked to traditional data bases containing ordering and shipping information. On the factory floor, several work cells may be integrated with a material handling system to create an FMS. In the factory of the future, these continents of automation will be integrated into worlds of automation that wfl] eventually encompass not only entire factories, but also entire corporations. Because of the volume of data and complexity of decisions needed for full integration, a hierarchical structure is the only feasible way to achieve it. A hierarchical structure has certain implications for the ar- chitecture of CIM systems. Data use and decision making must occur at the lowest levels possible. Only certain summary data will be passed upward in the hierarchy to be used in reporting the factory's state and in statistical trend analysis. Thus, an in- formation and decision hierarchy is needed that practices man- agement by exception. If additional information is required at upper levels, it will be requested. If local decision making can- not resolve a conflict, a decision wiB be requested from above. Conversely, management decisions may be communicated almost instantly throughout the system for rapid compliance. The hierarchical structure further implies the use of district uted data bases and distributed processing. A mainframe com- puter may be the host computer to the factory of the future. Connected to it will be an array of minicomputers, one level down in the hierarchy, each acting as host controller to an intermediate- level manufacturing system. A mix of local and centralized data storage will be appropriate for each computer. Below the m~ni- computers will be microcomputers acting as cell controllers, graph- ics work stations, or executive work stations. The elements of this hierarchical structure can be thought of as subsystems, categorized by the role they play, although it must be remembered that categories may overlap considerably. Most subsystems of manufacturing fall into one of the following broad categories: (1) information and communication, (2) inte- gration of processes, or (3) resource allocation. Note that each
122 category cuts across traditional manufacturing boundaries. An information-communication subsystem, for example, could include a network that permits the geometric part data stored in the CAD data base to be transformed through GT techniques into an actual process plan and then into robot and NC programs communicated to the factory floor. The data would then be transformed and com- municated for process scheduling and material handling, right up to the delivery of the finished product. Information-oriented subsystems include traditional manage- ment information system and data processing roles. These sub- systems will be able to expand to include geometrical data from CAD systems, material and process data from GT coding, parts- in-process data, and order and inventory data. Information sum systems will have analytic capabilities by which the data can be massaged for quality control and trend analysis. Data retrieval will be easier for operators and decision makers through the use of new query languages or programs that wiD allow nonexperts access to complex data. Most, if not all, manufacturing subsys- tems have strong information and communication functions, even if they are primarily process or resource oriented. Manufacturing subsystems on the factory floor will integrate traditional manufacturing processes by coupling and controlling previously separated processes and by carrying out computer- generated process plans. At the lowest level, this will involve data communication from sensors to a computer-controlled machine or robot. This provides the real-time adaptive control necessary to improve the work quality and throughput of individual stations. At the next level, factory floor manufacturing subsystems can in- tegrate several processes, such as an NC machining station, an automated inspection station, and the robot which services them. In this example, the coordination is supplied by a computer which controls the work cell. The process plan is downloaded from a computer, which may be in the CAE area, to the work cell con- troller, which coordinates the processing by the machines in its cell. With automated inspection and data collection, the process plan may be modified to eliminate defects by responding in real time to tolerance changes. At yet a higher level, work cells are integrated into an EMS so that an automated scheduling system
123 can assign a part in process to the next available work cell that can perform the necessary operation. This allows a system with fewer parts in process, shorter queues, fewer holding areas, and much more efficient use of floor space. Resource aBocation subsystems span a broad scope from small-scale material handling systems serving individual work cells to broadly implemented systems that monitor and control inven- tory, schedule work, and allocate materials to the factory floor on tight schedules. Automated material handling systems can be integrated into work cells and families of work cells to produce a powerful FMS. In turn, the FMS can be linked to production plan- ning and capacity planning systems to form the fully computer- integrated manufacturing systems illustrated in Figure A-1. One of the most important reasons for implementing smaB- scale subsystems of manufacturing now is that they can be suc- cessively integrated into these higher order entities, CIM sys- tems, that benefit from the synergy between operational pro- grams, product data, and process data. CONCLUSION All of these advanced manufacturing technologies, from ma- terials and machine tools to the subsystems and CIM systems, provide the ability to perform traditional manufacturing tasks in a highly advantageous but nontraditional manner. Many of the individual technologies and subsystems of manufacturing can be implemented today and, in fact, must be implemented soon for a manufacturer to remain competitive. Real progress toward the factory of the future will take place through the higher-level in- tegration of these technologies. Although a handful of domestic manufacturers continue to make progress in implementing and in- tegrating many of the technologies described in this appendix, real barriers to full integration remain. Specifically, standards are critically needed for the definition and communication of part data. At higher levels, the need is for proven systems of hierarchical control and feedback and usable methods of automated classification of parts and processes (GT). Required at the highest level are the evolution of structured anal
124 ysis and design systems that include modeling and optimization packages, as well as intelligent user interfaces that can be used interactively by managers in read time. The technology Is here now or just around the corner. U.S. manufacturing needs far- sighted management and trained manufacturing engineers to put the pieces together. NOTES iMore detailed and comprehensive information on current de- velopments in metal matrix composites can be obtained from the National Research Council's National Materials Advisory Board. 2 The New York Times. November 17, 1985. Goodbye to Heavy Metal, p. F-1. 3 this section is based on Research Briefing Pane! on Ceramics and Ceramic Composites. 1985. Research Briefings 1985, pp. 60 71. Washington, D.C.: National Academy Press. 4Advanced Ceramics Technology. Pp. 23-24 in Technology Today, March 1986. Committee on the Status of High-Technology Ceramics in Japan. 1984. High-Technology Ceramics in Japan, p. 30. Wash- ington, D.C.: National Academy Press. 6 Ibid, pp. 24-25 and 33-34. Gives market growth projections in Japan, as well as different applications given several price sce- nar~os. 7Interestingly, the initial patents for AGVs were held by a U.S. firm, the Barrett Corporation. However, consistent with the European dominance in material handling equipment technology that has existed for many years, the Barrett Corporation was ac- quired by a European firm, the Mannesmann Demag Corporation. This section is based on Thompson, Brian, 1985, Fixtur- ing: The Next Frontier in the Evolution of Flexible Manufacturing Cells, CIM (March/April):1~13. 9A good description of current developments in micromechan- ics can be found in Brandt, Richard, 1986, Micromechanics: The Eyes and Ears of Tomorrow's Computers, Business Week (March 17):88-89.
125 '°Smith, Donald N., and Peter Heyler, Jr. 1985. U.S. Indus- trial Robot Forecast and Trends: A Second Edition Delphi Study. Dearborn, Mich.: Society of Manufacturing Engineers. iiCommittee on Army Robotics and Artificial Intelligence. 1983. Applications of Robotics and Artificial Intelligence to Re- duce Risk and Improve Effectiveness, p. 58. Washington, D.C.: National Academy Press. i2 Committee on the CAD/CAM Interface. 1984. Computer Integration of Engineering Design and Production: A National Opportunity, p. 11. Washington, D.C.: National Academy Press. i3Shunk, Dan. Integrated Cellular Manufacturing. Paper pre- sented at the WESTEC conference, Los Angeles, March 1983. id Computer Integration of Engineering Design and Produc- tion, p. 50. i5Ibid., p. 17.
