Architecture and process: the role of integrated systems in concurrent engineering introduction.

The last decade of the twentieth century will be critical for business enterprises driven by engineering. Companies are being compelled by unprecedented economic, social and technological change to improve cost and efficiency throughout the life cycles of ther products. The technology leaders in electronics, aerospace, automotive and other discrete manufacturing industries have reached a ceiling on the amout of productivity they can gain from the use of computerized support systems and applications for discrete tasks. The next important stage of computerization for them is the integration of these systems and applications into distributed networks capable of supporting concurrent engineering.

Concurrent engineering methodologies permit the separate tasks of the product development process to be carried out simultaneously rather than sequentially. Product design, testing, manufacturing and process planning through logistics, for example, are done side-by-side and interactively. Potential problems in fabrication, assembly, support and quality are identified and resolved in the design process. The result is the ability to get high-quality products to market faster and at lower cost (see Figure 1).

Competitive pressures and problems

While the market for new engineered products is expanding rapidly throughout the world, this expansion brings with it a whole new set of business challenges. There are more competitors today. Markets are not homogenous; they are fragmented into increasingly focused niches requiring greater flexibility in the product mix, shorter manufacturing production runs and--above all -- higher quality. Technology is at once part of the competitive edge and, for the company struggling to better a rival's product design or manufacruing processes, a competitive pressure. Today, while both are important, the superiority of a product's technology simply does not carry as much weight as time-to-market.

Ralph E. Gomory wrote in the Harvard Business Review, "One cannot overestimate the importance of getting through each turn of the (product) cycle more quickly than a competitor...Even if a company starts out with an inferior product, it can overtake the industry leader if it has the capacity to turn out a new line six or 12 months more quickly."

This is well-illustrated in the automotive industry. Japanese auto makers captured a major share or the worldwide market by utilizing: statistical process management methodologies developed by Deming, Taguchi and others; teamwork with an emphasis on everyone's responsibility for quality; design for a manufacturability; and continuous improvement rather than major technical breakthroughs. The MIT Commission on Industrial Productivity found that the Japanese can get a new model to market in 3.5 years, a year and half faster than their American competitors can.

A worldwide overcapacity in manufacturing is putting severe pressure on the auto makers' profit margins. New product development is capital intensive. The program cost for a new automobile averages $2 billion. This includes design, engineering and tooling. The break-even point is the number of cars a manufacturer has to sell to overcome this fixed cost. To lower the breake even point, the manufacturer has to sell more cars or lower the fixed cost. High-volume manufacturers are moving to more flexible manufacturing and smaller lot sizes to attack niche markets. Partnerships between companies and acquisitions are increasing. More important, the role of engineering in reducing fixed cost is being vigorously explored.

Electronics manufacturers are also experiencing enormous competitive pressures. In spite of explosive growth in worldwide markets, it is getting harder to suceed from a business standpoint. The market for semi-custom ASIC (application-specific integrated circuits) chips in telecommunications, computer networks and automobiles is stronger than ever. In computers, electronics and consumer electronics, however, technical advances have driven cost and quality up while competition has driven prices down. The semiconductor industry is the best example of the effects of cut-throat competition. The woldwide market for Dynamic Access Memory (DRAM) chips, a commodity product, was $1.53 billion in 1987. It was projected to grow to $10.7 billion in 1991. Yet due to excess capacity and price competition, Texas Instruments Chairman Jerry R. Junkins told Business Week that 1988 and '89 were"...the first time the industry earned a cumulative profit in memory chips."

Problems in defense and aerospace

The aerospace and defense electronics industries have all of the problems of the automobile and electronics industries. These problems are magnified by capital costs in the billions, high risk, tight regulation and product cycles measured in years rather than in months. Defense Science reports that in the 1980s, the complexity of the product and manufacturing process technologies involved began to add their own special burden: "...new technology availability was driving defense systems into new, uncharted and extremely difficult areas, particularly from the manufacturing (read reliability) viewpoint. It was almost invetable -- in the absence of new industrial disciplines of some kind -- that things would fail, not work down at the 'soldier level.'" The use of more and more microelectronics, software development and maintenance problems, and complex new materials technologies have been three of the major complicating factors.

Consolidation and "teaming" are also characteristic of these industries. In commercial aviation, the number of large-transport integrators has been reduced to five: Boeing, McDonnell Douglas, and Airbus Industrial [a European consortium], deliver more than 95 percent of the aircraft. Companies like West Germany's Fokker and British Aerospace make smaller transport aircraft. They depend heavily, however, on technical and business alliances with an army of components subcontractors and suppliers. On aerospace/defense contracts, as many as 20 contractors and subcontractors is not uncommon.

The problems of complexity, cost overruns and reliability have become so serious that in the U.S. the Department of Defense has mandated a series of initiatives and standards designed to enforce improvement in the engineering, documentation and manufacturing processes. The most far-reaching DoD initiative is CALS (computer-aided logistics and support).

CALS provides a framework of standards for inter-operability essential to support concurrent engineering. For this reason, its influence is spreading beyond the American defense industry and into other discrete manufacturing industries worldwide.

Shifting the cost emphasis

MIT productivity study found that, "Prowess in research does not lead automatically to commercial success. New ideas must be converted into products that customers want, when they want them, and before competitors can provide them. And the products must be made efficiently and well."

Making a product "efficiently and well" starts with design engineering. Design engineering is the most important influence on product life cycle cost. In its report, The Role of Concurrent Engineering in weapons System Acquisition, the DoD cites a series of studies by Westinghouse Corp. Westinghouse found that by the time the concept for a new product has been defined, 20 percent of its total life cycle cost has been determined. By the time a prototype design has been developed, 75 percent of its costs have been committed. Only 10 percent of the product's cost can be influenced once it has been turned over to manufacturing. Recognition of this fact has shifted industry's cost containment attention from manufacturing to design engineering. Chrysler's Vice Chairman Gerald Greenwald says that in order to survive in today's auto market, "...yoy keep your quality up, you keep new products coming, and you work like hell to be a low-cost producer." Design automation has played a major role in keeping new products coming for the last 30 years because it gives engineers the tools to experiment with more sophisticated product designs and to perfect them through many iterations of design and analysis. Many products are, in fact, too complex to design manually. An integrated circuit engineer can design an average of 10-15 transistors per day without computerized tools. Current VLSI designs use upwards of four million transistors per IC. A 1-million transistor VLSI chip would take 50-75 man-years to design. Design automation will continue to be the largest technical application for computers, reaching $13,457 million in 1993. (Dataquest, Technical Computing Systems: Applications Analysis and Forecasts.)

However, the proliferation of incompatible CAD systems has created barriers to design process integration. Removing these barriers is so important to the bottom line that General Motors and its information systems arm, EDS, have undertaken a multi-million-dollar project called C4. The company is standardizing a specifc set of computer platforms and applications and is writing custom software to provide integration capabilities where open systems standards do not go far enough to meet the corporation's needs.

In the majority of companies, design engineering deparment work independently of manufacturing. But that is changing. Concepts such as design-for-manufacturability and design-for-assembly are being introduced into design engineering groups to ensure product quality at the least expensive stage of the life cycle. The further downstream an error or design flaw gets, the more expensive it is to correct with engineering change orders.

More protoype means higher costs and longer product inroduction delays which, in turn, trim profit potential (see Figure 2).

Manufacturing groups also have automated support tools for planning (CAPP), analysis and process simulation. But when these tools cannot share data and communicate with design automation systems, their impact on the life cycle is sharply limited. The early involvement of manufacturing engineers in the design process can shave months off time-to-market and improve overall product quality and cost. For example, reducing the part count in a product can yield cost savings that ripple throughout the entire production system, affecting manufacturing efficiency, reliability and serviceability of the product itself. The goal of concurrent engineering is 100 percent quality. To reach that goal, product design teams should include, for example, design engineering, software engineering, manufacturing planning and process engineering and other important product constituencies within the organization and throughout the extended enterprise (see Figure 3).

Essential components and benefits

The first essential component of a concurrent engineering environment is this ability to support teams. The DoD study researched concurrent engineering projects at Aerojet Ordnance, AT&T, Boeing, Deere and Co., ITT, McDonnell Douglas, Hewlett-Packard and other companies. It found that just giving decision-making responsibility to multi-function design teams helped resolve problems and improved processes and development times.

The second major component of concurrent engineering is a systems engineering approach. This refers to the use of analytical and statistical methods to improve processes. It also refers to the ability to weigh design, analysis, testing and design trade-offs among the various components and subsystems that can contribute to the success of a complete product. Manufacturing, logistics and support are as much a part of the product 'system' as its physical components. Cross-training personnel to approach design decisions from a systems perspective streamlines the development process. According to MIT's Kim Clark, "People who once worked in separate domains of technical knowledge or in research activities that could be conducted at arm's length -- product engineering and manufacturing ramp-up, say, or materials development and assembly automation -- must now be integrated. The focus of integration is a scientific understanding of the manufacturing process."

At Boeing, 'design/build' teams are now part of the lexicon. At Xerox, combining product and process design under a chief engineer who is also responsible for pilot production has helped the company reduce the development time for new copiers by almost 50 percent. Computer support is the third critical component of concurrent engineering environments. CAD/CAM produces drawings. CAE handles analytical computations. Together they provide data for faster, less costly electronic simulation and testing of prototypes. Comprehensive design databases and solid modeling applications allow design visualization. The design geometry captured in computers becomes data for numerically-controlled machine tools in the factory. CAPP and manufacturing simulation assist the manufacturing engineer. Data communications allows members of teams to work together even though they may be widely separated by geography. The benefits of concurrent engineering are leveraged by the degree of integration among these systems.

The DoD found that the application of concurrent engineering techniques and computer support improved the quality of designs and reduced engineering change orders by an average of 50 percent for all the companies studied. Boeing reduced its product development cycle time 40 to 60 percent. AT&T reduced the process time for a new microprogrammed digital switch by 46 percent. Deere and Co. reduced the product development time for construction equipment by 60 percent, its manufacturing cost by 30 percent and its scrap and rework costs by as much as 75 percent. McDonnell Douglas reduced rework costs 29 percent, scrap costs 57 percent and nonconformance by 38 percent "...through a corporate renewal effort that incorporated improved teamwork, better computer support and renewed emphasis on process controls." In addition, the companies studied reported reduced leadtime for creating bid proposals, reduced costs in design phases, during fabrication, manufacturing and assembly, parts reduction and improved inventory control These results came from projects at the local level. "But the really impressive savings, (hundreds of millions of dollars)," according to the DoD, "remain largely unrecognized because they result only from improvements in the larger 'systems' over which only top management has control."

A model for integration

Improvement in the larger systems -- company-wide, enterprise-wide--requires an integration architecture that will assist users in managing complexity.

Standards for data communications, database structure and access methods, data formats, windowing, graphics, document exchange and user services are essential. The U.S. Department of Defense Computer-Aided Acquisition and Logistic Support (CALS) Initiative was "...proposed in 1985 to establish a foundation for full communication between computer systems...The initial thrust of CALS has been to publish a series of implementations of existing national and international standards that allow the interchange of engineering and logistical data such as engineering drawings and technical manuals," according to Winner.

CALS includes among its standards: program-and device-independent specifications for storing graphics and raster data, specifications for the exchange of text, graphics and geometric data, database management and control, and communications.

Phase of CALS focused on substituting electronic documentation for the mountains of paper that flow between defense contractors, subcontractors and the government. Documentation is critical in the product life cycle. McDonnell Douglas reported reducing product cycle times 20 to 25 percent by using CALS digital data instead of paper methods. CALS Phases 2 and 3 are moving in the direction of enforcing data access and management standards in support of concurrent engineering practices.

Of particular interest, at this time, are the PDES (Product Data Exchange Specification) in the U.S. and its International Standards Organization (ISO) counterpart, STEP (Standard for the Exchange or Product Data). PDES/STEP is an evolving standard which will replace the IGES (Initial Graphics Exchange Specification). PDES/STEP is more useful than IGES; in addition to providing a common exchange format for data, it also provides more complete product information and powerful database access using expert systems technology.

Another important evolving standard is PEX. PEX is an extension to the X-11 windowing standard and PHIGS (Programmer's Hierarchical Interactive Graphic Standard). X-11 is a common graphics-oriented user interface for applications running on different computers in a network. PHIGS is the standard for 3D graphics programming.

PDES/STEP and PEX will facilitate and simplify the sharing, accessibility and maintenance of engineering data in networked environments which includes many different kinds of computing and applications resources.

Computing systems include hardware, operating systems and networking. Users need a range of platforms -- terminals, personal computers, workstations, servers, mainframes and supercomputers. They need to be able to share file systems among them. They need to provide network support for them, locally and anywhere in the world. The model for the way organizations implement computers is changing to reflect the need for higher levels of resource sharing, communication and integration. Digital Review reports research that indicates "the computer industry will see wrenching transitions in the next five years...leaving a wide open market for networked computing environments based on the client/server model."

The important standards for inter-operability in the client/server environment are: ISO/OSI. The International Open Systems Interconnect specifications for data communications and the de facto RCP/IP standard for local area networking in the UNIX environment; SQL (Structured Query Language) for relational database access; and POSIX, the emerging standard for an open operating systems interface.

From the user perspective, applications are the most important element in the integration model. Today, users have an enormous investment in software and data tied to specific applications in technical documentation and electronic, mechanical and software engineering. Much of it is provided them by leading independent applications developers. Many critical applications programs. especially in the aerospace/defense electronics and auto industries, are proprietary. Important standards in tying together applications are: X-11; and the proposed ATIS (Application Tools Integration Services) standard.

Networking provides the infrastructure

Given the complexity of today's engineered products and the international scope of business, marketing, research, engineering and manufacturing operations, effectively sharing information in a concurrent engineering environment starts with a communications infrastructure.

The Japanese firms are able to achieve almost complete overlap or product and process engineering, for example. To make this work, an enormous amount of information must flow between the product designer and the process designer. Much flexibility is needed as well, because of the inevitable changes on both sides as the design proceeds. (Detouzos)

Electronic mail, electronic conferencing and electronic 'bulletin boards' (or notes files) facilitate information flow throughout the product life cycle. Experience has shown that they eliminate the barriers of distance and time zones. They foster an open exchange of ideas and a cooperative, multi-disciplinary approach to problem-solving which cuts through the red tape of traditional organizational heirarchies.

George Fisher, CEO of Motorola, told Harvard Business Review that, "Another set of important linkages -- as important as any taking place within companies -- are between information systems of different companies. The electronic data interchange (EDI) systems that are being established between customers and suppliers are becoming two-way interfaces. The standards being established for them are also two-way." Fisher cites the EDI links between Motorola and Hewlett-Packard, a major customer. "In fact, it's already happening," he says. "Demand from HP factories can drive the forecasted materials planning systems in some of our semiconductor factories."

The MIT industrial productivity study found that "The vertical linkages in an economy which connect a firm with its suppliers below and its customers above can be conduits not only for incoming materials and finished products but also for technological innovation and other developments that enhance productivity." The study cites the success of the German machine tool industry in which customers ar closely involved in the engineering process and set higher-than-average standards for quality. The result is overall improvement in product cost and performance.

Data and process management

A network architecture based on standards including ISO/OSI. X.400 for electronic mail and EDI provides a highway for information. CALS standards permit data to be transported over that highway in formats common to CAD/CAM systems. Computers and people can communicate. But, for the most part, applications still cannot. They have their own unique data structures, their own user interfaces.

Therefore, in this environment which is meant to improve productivity through each turn of the product life cycle, data management and process management are critical bottlenecks. Decisions are being made, documents are being produced, and parts are being manufactured using out-of-date information because revision control is out of control. Time is being wasted searching among multiple

databases for information. There is a lack of security. All of these problems result in delays in product development.

Process management enforces adherence to standard, optimal methodologies -- workflow management. It also ensures that automated tools are used in the proper sequence and that they utilize complete and accurate data--configuration management. Implementing an effective process management program requires careful modeling of the user's unique product development data flow in advance of planning for applications integration.

To manage the growing output of automated systems and many different applications using manual techniques one cannot adequately control and disseminate today's volume of computer-generated data. In the aerospace/defense industry, according to the MIT study, "Managing engineering information is difficult in the presence of integrated and continuing participation of multifunction teams in the design or product and processes. The requirement for flexibility to evolve the engineering, manufacturing and support processes place an added burden on information management. This burden is increased by two factors: widely distributed teams and the size complexity of weapons systems...open and continuous communication between customer and vendor. All of this indicates a need for evolvable, tailorable, inter-operable, secure, distributed and high-performance enterprise information management systems, starting with engineering and production."

Computer-aided software engineering (CASE) is one are in which applications integration has proven its value. "Experience with CASE has shown that no one tool is going to have the kind of impact that software developers need. While individual tools can increase productivity on the order of 10 to 15 percent, the portions of the development process that remain un-automated become the critical bottlenecks that constrain overall gains...The maximum benefit of CASE is realized when all components are fully integrated into a seamless environment -- one that appears consistent to the user and has efficient mechanisms for sharing data among individual tools.

Since the development cost for providing all the tools required is extremely high, open architectures that allow the vendors to cooperate by interfacing their respective products will a growing trend in CASE," according to an article in CASE Outlook.

ATIS (Application Tools Integration Services) is a proposed specification for an applications integration framework. It has been submitted to CALS, PCTE+ (the European Portable Common Tool Environment) and IRDS (Information Resource Directory Service) committees. IRDS is a set of standards to provide a common data dictionary for database systems in a distributed, networked environment. The purpose of ATIS is to support open integration and applications portability. It is essential to provide enterprise-wide data and process management.

During the previous decade, the engineering market was driven by the need to improve the performance of discrete applications and workstation sales exploded. A research report developed by the industry research firm Dataquest states, "The rules of the workstation market are changing. Since the early 1980s, workstation vendors were primarily concerned with performance; the machine with the biggest MIPS and MFLOPS performance at the lowest price was the winner of the workstation game. Today, however, the rules are different. Workstation vendors must learn to anticipate the demands of their customers. Service, software and support are becoming more and more important issues to their customers."

Concurrent engineering holds out the promise of huge rewards to companies willing to undertake the effort -- shorter development cycles, higher quality products, lower costs and the ability to compete more effectively in dynamic world markets. The organizational and computing methodologies required are, however, complex. Long-term commitment and leadership must come from the top of the organization. Success will be achieved through incremental stages of integration starting with the design engineering department and spreading out -- via stable, powerful integration architectures--to include the extended enterprise.

For futher reading

Avishai, Bernard and William Taylor, "Customers Drive A Technology-Driven Company: An Interview with George Fisher." Harvard Business Review. Nov/Dec 1989.

Bartimo, Jim, "TI Bets Most of Its Marbles in Chips," Business Week. Jan 29, 1990.

CAD/CAM Industry Overview. San Jose, CA: Dataquest, Inc.

"CAD Environments for the Software Life Cycle," CASE Outlook July, 1987.

Clark, Kim B., "What Strategy Can Do For Technology." Harvard Business Review. Nov/Dec 1989.

Depompa, Barbara, "Computer Services and Software Will Grown in Demand in 1990." MIS Week. Jan 8, 1990.

Dertouzos, Michael L., Richard K. Lester, Robert M. Solow and The MIT Commission on Industrial Productivity. Made in America: Regaining The Productive Edge. Cambridge, MA: The MIT Press, 1989.

Gomory, Ralph E., "From the 'Ladder of Science' to the Product Development Cycle." Harvard Business Review. Nov/Dec 1989.

Gross, Richard C., "shortfall in Technical Manpower Next Crisis." Defense Science. Dec 1989.

Grygo, Gene, "Client/Server Model Signals 'Macro Change' for Industry." Digital Review. Dec 18, 1989.

Hall, George M., "Know-How Versus Know-When in Systems Integration." Information Strategy: The Executive's Journal. Winter 1990.

Hammonds, Keith, "How a $4 Razor Ends Up Costing $300 Million." Business Week. Jan 29, 1990.

Harvey, David, "TQI--A Revolution Spurned." Defense Science. Dec 1989.

"Industry Outlook: 1990," Business Week. Jan 8, 1990.

Technical Computing Systems: Applications Analysis and Forecasts. San Jose, CA: Dataquest, Inc.

Technical Workstatins: Industry Overview. San Jose, CA: Dataquest. Inc.

Winner, Robert I. et. al. The Role of Concurrent Engineering in Weapons System Acquisition. Unclassified report prepared for the Office of the Assistant Secretary of Defense for Production and Logistics. Alexandria, VA: Institute for Defense Analysis, Dec 1988.

John Izuchukwu is the manager of aerospace strategy at Digital Manufacturing Cluster, Digital Equipment Corp. He is a member of IIE and is pursuing a doctoral degree in industrial engineering from Northeastern University, Boston.