Enabling Technologies for Unified Life-Cycle Engineering of Structural Components (1991)

Information flow has been characterized as being dominantly a one-way flow: from design to manufacturing to support. In some cases the materials community is not included at all. Life-cycle information flow from manufacturing and support back to the product and process designers is incomplete and slow. Many designers have new assignments before the product passes into manufacturing, and the product life cycles are so long that designers are almost certain to have new assignments before their designs attain maturity. Useful feedback to the designer on manufacturing problems for current products is also limited, as is feedback on current support problems.

At present, the question of manufacturability is a complex, multidisciplined one. Complete information on the suitability of a product for manufacturing includes costs, conformance to process capabilities, assemblability, fragility, sensitivity to the manufacturing environment, testability, and many more. The designer needs information about each facet of production to ensure an appropriate product design. As mentioned, this information is not generally available in the form of feedback on current similar designs. Also there is an incomplete set of design assessment tools to allow a designer to test the design for manufacturability or for supportability. Integrated Computer-Aided Manufacturing Definition (IDEF) models are becoming the standard for describing information flow requirements for integrated manufacturing systems. This modeling method is expected to attract additional users in the near future. The new emphasis on looking at information flow within a manufacturing enterprise will provide a sound basis for integrating ULCE technologies and practices.

Similarly, the question of supportability is complex and requires information from many disciplines. Designers need to consider cost, reliability, repairability, environmental sensitivity, use beyond the original operating envelopes, etc. Currently, only a limited amount of information can be made available to designers to aid them in making decisions that will enhance the supportability of the product. The best information comes from direct interaction with support personnel, but its utility for new designs is limited. Feedback on failures and repairs is quantitative but without supporting qualitative narrative and item history. Modes of damage and battle threats are not generally communicated from operations back to the designers. Even basic information on use and the operating environment is not fed back to the designers.

It is difficult to notify designers of lessons learned or new alternatives. Often the information to improve designs exists, but it resides in different data bases, with different access methods, and with different contexts for understanding.

Computer-based information systems can be developed which will provide better feedback to designers but other issues related to security and accountability will arise. For example, information security for traditional methods (paper) is reasonably complete, but computer-aided design systems have not yet demonstrated that they can provide appropriate protection. Also, current technologies do not provide audits to ensure that required procedures have been followed and that required approvals have been obtained, especially when information crosses organizational boundaries.

Current data systems require prohibitive software modifications, incurring high cost and delay, in upgrading to meet new requirements. They reflect current functional and company organizational structures, and in many cases modifications lag and even prevent organizational improvements. They do not easily permit the addition of a new arbitrary data type—e.g., vision data for inspection cannot be added to a numerical-control program file.

The current state of the art in artificial intelligence technology supports one main application area—expert systems. These applications are generally rule-based systems. The rules represent knowledge of the problem domain with which the system is concerned. In general this knowledge is obtained directly from human experts. There are several hundred applications in current use, with a focus on stand-alone manufacturing systems. An Air Force project to develop meta-level knowledge for computer-integrated manufacturing is underway and is coordinated with an object-oriented programming project to improve knowledge-representation capabilities. Rule-based expert systems are beginning to be used to process engineering notes for comparison with design rules. These systems may be used to communicate from design to manufacturing, but there are no general methods for feedback to the rules base to reflect new manufacturing knowledge. The role of artificial intelligence technology in providing advanced design capability has been discussed by many authors (Mostow, 1985; Lakin et al., 1989; Ulrich and Seering, 1988).

The International Graphic Exchange Specification (IGES) is an industry standard for communicating the geometric portion of a design definition, but there is no standard descriptive language for communication of the nongeometric and manufacturing process specifications associated with a design. Electronic communication of design definition is difficult, but several government-initiated information exchange programs provide encouraging progress—Geometric Modeling Application Interface (GMAP), Product Data Definition Interface (PDDI), as well as Department of Defense (DOD) Standards, and American National Standards Institute (ANSI) (14.26M) Electronic Data Interchange.

The U.S. information processing market has supported numerous vendors of computer equipment both large and small. Unfortunately each vendor has chosen to develop systems which are vendor-specific. Hence, current computer applications are not readily transferable to other hardware or operating systems, many depend on a particular data-base manager, and some are dependent on a particular display terminal. However, this is slowly changing. Major computer vendors have announced plans to support applications across a wide variety of operating systems, data base managers, and communications protocols and application architectures are being announced that will allow transfer of applications without major revisions.

Advanced function engineering workstations are available that provide a significant improvement in computer power and graphics display for designers. Research is in progress by the Air Force to investigate the best architectures for parallel processing for ULCE problems. ULCE data exist in hierarchical, relational, inverted, flat files and other data base managers. Relational data base managers are available to link data from multiple paradigms to a common access facility. The Integrated Information Support System (IISS) provides access to data on a heterogeneous collection of data-base managers across a distributed environment. Other current activities that will aid the integration of data from diverse sources include the Integrated Design Support System (IDSS) and the Integrated Manufacturing Data Administration System (IMDAS).

In the future, design systems will include a complete set of design assessment tools and access to current manufacturing and support information on similar designs. There will be a more complete characterization of the production and operating environments, since future products will include embedded sensors that will generate feedback from manufacturing and support to engineering design. Integrated information will be available to designers and support personnel. In the ULCE environment, information on current and predecessor products and processes will be available to designers as they work on new products. Future development cycles may be shorter, so designers will have better opportunities to track their work into production and to obtain firsthand information on manufacturability. The current serial design process will be replaced by concurrent processes by providing interactive access for many disciplines (including reliability, supportability, manufacturing process) to the design definitions as they are developed. Future systems will permit a complete audit trail of decisions and authorizations and, to some extent, will ensure their occurrence.

Many of the concepts, analysis, and research to support computer-integrated manufacturing (CIM) are part of the elements required for ULCE. Key CIM activities are taking place in Europe under the European Strategic Program for Research and Development in Information Technology (ESPRIT) Computer Integrated Manufacturing Architecture Project, and in the United States at the National Institute of Standards and Technology's Advanced Manufacturing Research Facility. Within the next 5 years the CIM architecture developers will have carried out a first complete pass at developing a generic open system architecture for manufacturing enterprise information. They will have identified the major system objects and will be ready to prepare standards for the International Standards Organization.

In the next 5 years artificial intelligence systems will be extended to include abilities to infer rules from data, improve themselves, monitor and maintain self-consistency, accept and understand human speech, scan and understand unstructured text, and understand and communicate graphics representations. Neural nets may offer an opportunity to understand complex phenomena from experimental data and to supplement missing or faulty data.

Extensions beyond 5 years are difficult to project. In the future, complete standards for electronic exchange of product definitions will be in place. Telecommunications among contractors, vendors, and customers will be commonly used and will include full-motion video facilities. Future data systems will include the ability to add new data types (telemetry, video,

visual images, comments) as necessary to data bases. Conversion tools from current paper methods to digitized records will be available to automate the transition and to establish historical references. Highly improved methods of communication will be employed to transfer design and product quality data among contractor, customer, logistics organizations, vendors, and product users. There will be shared learning across the interfaces of these functions. Data access will be routine across diverse systems, and integrated systems will accommodate natural language, audio, video, and graphics. There will be a generally available data base of materials properties.

Information reference models will be in place to provide definitions for major components of CIM systems. These same models may be able to be modified with minimum effort to include the additional requirements of ULCE.

Future data systems will permit the addition of new data types to existing data bases as the need arises. Research to support this capability will continue during the next 5 years. Finally, there will be coordination and cooperation in the development of ULCE computer systems so that suppliers, contractors, and the government (vendors, manufacturers, and customers) will be able to communicate requirements, commitments, and information across their physical, geographic, and organizational boundaries.

As in CIM, ULCE concepts are based on the availability of unified data access across a manufacturing enterprise, including its suppliers and its customers. Data access in a heterogeneous environment of processors, communications networks, data-base managers, and user contexts without end-user awareness of the complexity of this environment is required. Current research will continue, as will an increased commercial emphasis to reduce research results to practice. Closer ties between research and commercial development will enhance the technology transfer. Additional personnel will become end-users of ULCE systems, and many of the current and future users will need to tailor general solutions to their particular operating environment, plant configurations, changing business and policies, and so on. They will need to accomplish this tailoring without the use of data processing and/or computer science skills that are different from their regular assignments. High-level commands and languages will make this user productivity possible. Software system prototyping will be in widespread use as a software engineering tool.

Lakin, F., J. Wambaugh, L. Leifer, D. Cannon, and C. Sivard. 1989. ''The electronic design notebook: performing medium and processing medium,'' The Visual Computer, 5, 214.

Mostow, Jack. 1985. "Toward Better Models of the Design Process," AI Magazine, 6, No 1, 44.

Ulrich, K. T., and W. P. Seering. 1988. "Function Sharing in Mechanical Design," Proceedings of the Seventh National Conference on Artificial Intelligence (AAAI-88), 342, American Association for Artificial Intelligence.