Information Technology and Manufacturing: A Preliminary Report on Research Needs (1993)

To turn design ideas into reality, production (e.g., material modification, assembly, testing, recycling, or information processing) takes place in a factory. Iron ore is converted into steel in a foundry; steel is converted into sheets, rods, and bars in a mill; these various forms of steel are fabricated into wire, screws, the structural support for buildings, washing machines, automobile bodies, and so on. Sand is transformed into elemental silicon, which is layered on a ceramic substrate to form an electronic circuit; these circuits are connected (or assembled) into a vast array of more complicated parts and devices such as radios, televisions, and controls for washing machines, automobiles, and spacecraft.

Sophisticated information technology is enabling increased automation of traditional production techniques, as well as new techniques such as stereolithography and material deposition. This chapter presents a vision of the shop floor and production systems of a factory in the year 2010 in light of these changes, and it describes research needs to achieve the vision.

In moving toward 21st century manufacturing, information technology will be used increasingly to make real-time determinations of, for example,

• How best to respond to customer demand,

• Which process flow path to follow and when,

• What material requirements exist,

• What optimal inventory levels to maintain,

• How manufacturing is performing, and

• What the project and unit cost will be.

Information technology will facilitate better communications and controls among contributors to the manufacturing process. Within factories, machine-to-machine communication will be facilitated by standard protocols or interfaces for control, diagnostic, and repair information; utilities; raw material feed; in-process handling; finished product disposition, and so on. Information will be increasingly embedded in parts and products and read by material-handling and processing equipment, further automating the flow of materials and work in process. These and other factors will make possible the instantaneous product and process flexibility that will be necessary to compete and prosper.

Material-handling devices and materials used are also expected to change. The material-handling devices in the envisioned production environment will extend from the massive to the nano level; manipulators will exist to operate on microscopic parts and assemblies. Increasingly, the materials used will be synthetics, composites, and ceramics.

Physical shaping will take place by means of such processes as water jets, laser cutting and welding, and thixomolding,¹ and new processes will be more flexible and characterized by less use of hard and soft tooling. The envisioned shop floor and production system environment creates a number of research opportunities. These are listed in Table 3.1 and discussed in the remainder of this chapter.

Essentially all modern machines, test stands, and instruments use embedded computer controls that cost more and deliver less than common personal computers. Equipment controls as delivered from equipment suppliers today are typically lacking in reliability, flexibility, robustness, and ease of integration with factory systems.

Information technology can play a vital role in improving the routine control functions of machine tools, robots, automated guided vehicles, and many other basic processing machines on the factory floor. Equipment in the envisioned production environment will be instructed by common (or at least communicating) controllers. This equipment can be thought of as a collection of “intelligent agents” reshaping raw stock into finished products.

The controllers will be object-oriented, self-revealing, and model-based. As a result, factory and enterprise software engineers will be able to integrate equipment faster and easier than possible today² and to use and control the equipment more flexibly. Also, such controllers will enable computer-aided manufacturing systems to orchestrate more carefully the interaction among all factory equipment and resources.

Research is needed to advance the state of control architectures and related software. For example, in contrast to today’s technologies, new control architectures and diagnostic software routines are needed to support the nascent field of intelligent sensors (discussed in the next section). Control architectures and software must also support communication with upstream functions, including computer-aided process planning and computer-aided design.

Research for open-architecture manufacturing should address:

• Appropriate operating systems, languages, data structures, and knowledge bases;

• Architecture and technology for shop floor equipment and data interfaces;

• Architecture for control systems;

• Design for repairability and the ability to work around equipment crashes, including diagnostic software;

• Better real-time control; and

• The human-machine interface to permit people to interact effectively in this environment.

Beyond research, standards and related assistance (e.g., from Sematech activity for semiconductor manufacturing) are needed to ensure consistency and interoperability of implementations, to coordinate and standardize open systems concepts in control systems, and to motivate system and equipment suppliers to use these mechanisms in their equipment. (See “Architectures and Standards” in Chapter 4.)

The manufacturing enterprise has long depended on sensory devices to provide realtime feedback about conditions of the process during manufacturing. Historically, sensors served only as production monitors. Increasingly, they will become active components of production systems, as integral parts of either a process or a finished product.

Examples of processes into which sensors will be incorporated include extrusion dies for temperature and metal flow control and surface finish, turning tools for thermal control to provide maximum life and load, and continuous processes for transmitting process

parameters. Sensors will interface so closely with processes that complicated interpretation of data now done externally will no longer be necessary.

Examples of finished products into which sensors will be incorporated include aircraft wing skins for dynamic analysis and control of the aircraft; shafts for torque readings; engine die castings for thermal and strain information; structure members for load and corrosion information; and tubing for providing information on pressure, temperature, or electromagnetic radiation.

Sensor-loaded products may be able to know where they are, how they are being used, when they have been damaged, or when they fail to meet a specified parameter. And they may be able to monitor their environmental impact and signal for containment or destruction. The new active sensors will change the way products are transported, tracked, monitored, and maintained at every stage of their existence—from creation to recycling to disposal.

The production environment envisioned for 2010 will implement real-time, dynamic scheduling systems, which reflect the external priorities of the manufacturing enterprise and balance them from moment to moment against circumstances prevailing in the plant and in the manufacturer’s supply chain. Conventional batch planning systems cannot recognize sudden changes in conditions generated by drifting machine capability, high scrap production, material shortages, or unplanned downtime. Such changes in condition often require costly crisis intervention. Furthermore, unnecessarily long queues form with conventional material requirements planning schedules, whereas the kanban method often results in empty queues and idle equipment.

With dynamic scheduling, by contrast, the overall utilization of material, labor, and equipment would provide maximum benefit by optimizing the whole system and modifying the operations at each cell, thus reaching levels unattainable with the best of present methods. The envisioned dynamically scheduled environment would continually track the status of jobs, cells, tooling, and resource availability. Through a communications network, each cell would have access to this information. What should be done next by any particular cell at any particular moment would thus be determinable at that cell and would be based on current conditions throughout the factory. The same communications network would also provide factory management with access to current production data.

Research on the following topics is needed to achieve such a dynamic scheduling environment:

• Dynamic shop floor models with high-speed recompute time and the ability to handle numerous variables;

• Real-time scheduling tools for the flexible factory and the distributed factory;

• Presentation tools to facilitate situation assessment and scheduling by the factory manager and operations team (see “Collaborative Technology and Computer-supported Cooperative Work” in Chapter 4);

• Visualization and human-computer interfaces to present the scheduling information;

• Multilevel understanding of large-scale systems (see “Complex Systems Theory” below);

• Means for identifying the relevant measures and quantifying the relative performance of the alternative systems; and

• Tools to support the brokering of priorities and obligations among cooperating entities, based on optimizing transportation, material handling, inventory, capital, and labor costs.

The envisioned production environment will incorporate an intelligent routing system, an emergent property of factories involving autonomous-operator workstations and random-path material handling. Each manufacturing cell in such a system knows the operations that it has been certified to perform and bids on work. Cells communicate their status to intelligent parts carriers through a communications network.

The intelligent routing system requires a process plan that is quite different from the computer-augmented process planning (CAPP) systems now in use. Typical CAPP systems, both variant and generative, have as their goal the selection of the best of alternative methods. An intelligent router, by contrast, will present acceptable methods (or routes), including one determined the best according to specified criteria (e.g., minimum cost, minimum speed, or maximum quality). All acceptable methods become functional alternate routings. Whenever an engineering change occurs, the routing is downloaded to all cells that have been authorized by the engineering change to bid for work on the part. In this sense, the intelligent workstation becomes an empowered agent (see “Architectures for Autonomy and Distributed Intelligence” in Chapter 4).

To realize intelligent routing systems, research is needed to identify appropriate interfaces among product design, product engineering, manufacturing engineering, and factory floor procedures as they will emerge in computer-augmented work groups, including human-computer interfaces. The immediate research focus should be on demonstrating the resilience of the intelligent routing system with respect to the vagaries of factory conditions.

To date, intelligence has been embedded in part carriers. The envisioned production environment of 2010 will support “smart part” routing, in which much intelligence is embedded in the parts themselves.³ Smart parts will be particularly useful for products whose production crosses building or enterprise borders, because the part itself carries the necessary production instructions and history in computer-readable media. As operations are carried out, associated instructions can be deleted or marked to indicate completion. Smart parts can also monitor and indicate performance. Thus maintenance requests can be triggered if the part senses its performance to be substandard. Further, planned maintenance regimens may also be recorded within the part, automatically triggering requests for maintenance.

Research is needed to identify practical, open standards for recording and communicating data among parts, assemblies, subsystems, production equipment, and maintenance equipment. Research is also needed into mechanisms for cost-effectively embedding the information and for ensuring access throughout the life of the part.

Manufacturing requires precise scheduling and control of numerous intricate processes. It requires substantial capital equipment (and, increasingly, software), labor, and management. To be competitive, a manufacturer must continually find ways to reduce costs, improve quality, and reduce the time needed to make the factory productive. And the manufacturer must meet these challenges in the face of constant change. Changes commonly stem from errors, tolerances, accidents, mistakes, and late deliveries (variance control factors), as well as from the planned introduction of new products and processes into an already-working environment (change management).

The traditional approach to determining how best to respond to change has been to experiment on or in the factory. The production environment of 2010 will support experimentation on a model simulation of the factory—a virtual factory—that can simulate operations without disrupting the actual operations.⁴ Both the physical and the information control and scheduling processes will interoperate and use the production software modules to give realism.⁵ The model will operate both in real time and in accelerated real time in order to test ideas quickly. To enable a range of factory personnel—from managers to operators—to make sound decisions, the model will be designed to operate at the appropriate level of abstraction for the task at hand without expensive and time-consuming tailoring from the user.

Although complete simulation of even a modest factory requires numerous models and may not be possible for a number of years, the time is ripe to pursue the virtual factory. Many of the supporting tools (e.g., a high-bandwidth communications network and factory modeling and simulation techniques) are available now or soon will be. Development of the virtual factory can build on large manufacturing simulations already implemented. For example, Boeing used virtual prototyping in its 777 project. And for semiconductor manufacturing processes, simulation often has proved to be as effective and valuable as running a real wafer through the actual chemical process.

An appropriate next step is the virtual production line, that is, a simulation of manufacturing process. The virtual production line involves simulation of individual tools and, more importantly, simulation of the integrated operation of the tools in production. A wide variety of individual, single-activity models are already in use in manufacturing. But the comprehensive integrated modeling of the manufacturing enterprise will provide new insight into the causes and elimination of scheduling bottlenecks and new strategic options, just as SIMNET did for military preparedness in the Gulf War.⁶

To achieve the virtual production line, and ultimately the virtual factory, modeling and simulation tools are needed that are as realistic as possible. Such tools will require a comprehensive model or structure for incorporating heterogeneous models so that the entire product realization process—from design through orders to multicountry manufacturing and distribution to customer delivery—can be tested, measured, and optimized.

Specific research issues include improving methods and tools for

• Efficiently managing large numbers of related data spread over many machines and locations,

• Specifying complex data relationships,

• Easing interoperability of design process tools,

• Encoding and decoding data, and

• Retrieving data (e.g., using human interface tools).

Manufacturing facilities of the future will not only be structured to handle one-of-a-kind customization, but also will have to be able to handle reasonable shifts within product families, and in some cases complete product shifts.⁷ Such abilities are the essence of the terms “agility” and “flexibility” as used to describe manufacturing of the future. In addition, manufacturing facilities will participate increasingly in virtual enterprise structures, made up of globally dispersed units that unite to carry out a specific task (e.g., to produce a specific product or product line) and disband when the task has been completed.

Reconfiguration of the production environment of 2010 will be achieved mainly through software but also through creative tool and product specification and design on each factory floor. With the trend toward increasingly customized product lines, systems to describe the product being ordered and check the validity of a proposed configuration are becoming increasingly necessary. At present, companies either develop product-specific systems or resort to artificial intelligence techniques to deal with configuration control. Both approaches have substantial deficiencies.

Research is needed to develop a universal product configuration language and methodology that could be used by both sales/marketing and manufacturing personnel to develop valid product configurations.

Research is also needed to assess the feasibility of programming languages to represent manufacturing operations in the same sense that design languages are expected to represent designs. Furthermore, research is needed to support the development of systems that simulate the operation of a given manufacturing configuration under a variety of conditions to enable a planner or scheduler to optimize a configuration (see section above, “Modeling of Manufacturing Systems”).

The most critical elements of any manufacturing system are the resources in the system; without them, product transformation would be impossible. Despite the importance of resource management, little research and effort have been devoted to creating generic resource representations. As a result, specific resource characteristics must be recreated each time a modeling activity is undertaken.

The envisioned production environment would support generalized models of systems for planning, analysis, and control. Such representations are necessary to produce models critical for both the planning and the operational control of manufacturing systems. Common representations and descriptions of resources are necessary in order to develop transferable (from planning to analysis to control) models and analysis. The models will be capable of describing

• Kinematic capabilities of the individual pieces of equipment, along with other process limitations;

• Processing capabilities of the equipment;

• Tool and fixturing capabilities associated with the equipment; and

• Contact space of the resources, along with other aspects of the system.

Specific research is necessary to develop representation schemata and models of manufacturing resources (tooling, machines, controller features, and so on) and their interconnection.⁸

In recent decades, many new physical processes have been invented to augment manufacturing production. Laser processing, water-jet cutting, and stereolithography are examples of relatively exotic, niche-oriented technologies. Electro-discharge machining (EDM) and chemical vapor deposition (CVD) are examples of technologies that were introduced only recently but are now common. Over time, a variety of new processes will emerge, and continued research is needed to feed that flow of technology.

New processes will be enabled by and will generate new requirements for information technology (e.g., for control technology). For example, the production environment of 2010 will include knowledge bases that describe the physics of such processes and provide for appropriate process control functions. Such knowledge bases will be able to model efficiently nearly any process that may be developed in the future. The research implications of such a system include software engineering innovations in areas such as data representation, incremental update, archival data, and data validity.

Manufacturing systems and other phenomena with numerous interrelated elements have been characterized as intractable, statistically random, or “chaotic.” Coupled partial differential equations, the traditional mathematical technique for describing such phenomena, are often inadequate for manufacturing applications because many elements of a manufacturing system cannot be characterized by partial differential equations.

New complex system theories are emerging that reveal some of these phenomena to be orderly and nonlinear. The theories have proved valuable in fields unrelated to manufacturing, such as genetics and artificial life.

Research is needed into the multilevel understanding of large-scale systems. The ability to analyze at the appropriate level of detail for the information sought is an important objective. This will require, among other things, the ability to model and simulate the behavior of such systems. Another promising approach to describing the behavior of large-scale systems is to view them as the cooperative action of autonomous agents acting in response to a performance function that emulates the actions of a free market. Research is needed to explore the validity of such an approach.