AUTOFACT '95:

A Report from Chicago

By David Blanchard/Editor

The award was announced at AUTOFACT '95, SME's annual conference and exhibition dedicated to automation and computer-integrated manufacturing (CIM). AUTOFACT '95 was held last month in Chicago.

The Beijing plant recently completed a four-year project to modernize its massive, outdated manufacturing facilities. The manufacturer wanted to increase NC milling machine development capacity and production efficiency while shortening delivery time, increasing quality, cutting costs, and supplying full service. Problem was, the plant was taking 24 months -twice as long as it should have been taking - for design, machining, assembly and test lead time. A CIM solution was chosen as the best path to achieve a level of global competitiveness.

The CIM program embodies an assortment of technological acronyms, including CAD/CAM (computer-aided design and manufacturing), CAPP (computer-aided process planning), MRP II (manufacturing resource planning), and FMS (flexible manufacturing systems). The Beijing CIM system links the entire manufacturing plant, including such tasks as monitoring new orders, scheduling, material use, shop floor control, finance, purchasing, and decision-making.

Obviously the modernization effort worked (awards aren't usually handed out for failures); in this instance, Beijing No. 1 Machine Tool Plant managed to eliminate 90 hours of the machining operation and reduced hand rework from 25% to next to nothing. One test run was able to produce a finished part from CAD data in two hours.

Toward Cellular Manufacturing

One of the characteristics of cellular manufacturing is that organizations are rethinking the once-popular "bigger is better" philosophy, and are choosing instead to "rightsize" into teams and groups. The objectives of cells, Miller explained, include: improved service; reduced inventory; reduced costs and overhead; improved quality; reduced space requirements; increased sales; and improved performance.

Miller cited a number of case studies, by way of illustrating both the real-world benefits to be gained by implementing cellular manufacturing, and the equally real-world disasters that can occur should management fail to support what amounts to a corporate paradigm shift. Education and communication, he stressed, are vitally important to keeping everybody on the same page in understanding the short- and long-term goals.

What Makes a System Intelligent?

Ananthraman explained that "process optimization solutions to real-world prediction, classification and control problems can be provided using hybrid systems developed using a combination of classical and novel methods." He proceeded to briefly explain the role each intelligent technology can play in manufacturing:

• "An expert system has a wide base of knowledge in a restricted domain (usually in the form of rules) and uses complex reasoning strategies to perform tasks at which a human expert is adept," Ananthraman noted. These expert systems (also referred to as knowledge-based systems) are used in such application areas as real-time process control, production planning and fault localization in industrial equipment.
• "A fuzzy logic system operates on the basis of rules as well; the rules used in this case are soft and adopt a `degree-of-truth' quality using linguistic terms such as light, moderate, heavy, etc. Any action or response resulting from a statement being true or partially true executes to a strength reflecting the degree to which the statement is true." Fuzzy logic has been adapted to the task of cruise control in automobiles and similar prediction and control applications.
• "A neural network-based system learns the correct control methodology simply by observing training data sets comprised of input-desired output pairs that are generated when the system is in operation." Application areas for neural networks include quality control, process control in the semiconductor and petrochemical industries, and optical character recognition.
• "Genetic algorithms attempt to mimic the mechanism of natural evolution to solve mathematically intractable problems in a wide variety of domains. They employ heuristically guided random search mechanisms to look for solutions in the system space," he explained. Genetic algorithms have been used for job shop scheduling, industrial optimization problems, and path planning and navigation for industrial robots. [For a more extensive review of the above-named intelligent systems, see the February 1995 issue of Intelligent Manufacturing.]

Adding Knowledge to Computers

• "Knowledge-based systems can multiply the know-how of skilled specialists by making it available to all engineers and designers through the CAD system network.
• "Knowledge-based systems can drive parametric CAD functionality, providing the ability to generate new designs automatically from changed input parameters with little or no manual intervention.
• "Knowledge-based systems can generate preliminary designs from customer specifications, create drawings, calculate material and labor costs, and produce bid estimates and proposals, often in a matter of hours.
• "Knowledge-based design systems can contribute significantly to reduction of lead times through efficient automated design, elimination of the need for physical mockups, and reduction of design iterations due to improved first-pass design quality."

Trico Products Corp. (Southfield, Mich.), a supplier of automotive windshield wipers, is using knowledge-based systems as their preferred tool to capture, retain, apply and evolve their best product development processes. According to Trico's Jeffrey Blake, "Knowledge-based systems have given Trico the opportunity to evaluate and redefine its entire approach to product development, not just product design."

Blake pointed out that these intelligent systems have made it possible for Trico to perform detailed analysis across a broad spectrum of product development activities, from conceptual design through cost accounting. The benefits have been impressive: In one instance, Trico was able to respond to a request in two hours, when previously it would have taken two days to run the cases, provide graphical data and reports, and make design recommendations. In another instance, a $60,000 tooling charge was saved when the intelligent system detected a design in violation of manufacturing rules.

A Smart CAD Tool

Brost and Peters demonstrated how HoldFast was used to design and build fixtures for finish-machining and drilling of a cast part for prototype fabrication and mass production fabrication. Pallet fixtures using HoldFast were designed for the vertical assembly of a personal cassette player as well as a glue gun.

Intelligent Process Control

MF/SmartMold provides an intelligent quality analysis and monitoring system, and consists of three major modules: Die Set, Process Window and Quality Monitor. The system, as Rowland and Leonard explained, "utilizes a series of rules embedded in an expert system to ensure the process parameters are moved from their initial start point to a point where good parts are produced." MF/SmartMold is in the final development stage, and early results indicate that it can predict primary defects such as short shot and flash.

Manufacturing on the Internet

Most of this information can be accessed via the World Wide Web area of the Internet, and include: "abstracts and texts of technical reports from manufacturing research centers; personnel directories of national laboratories, engineering departments and private companies; government reports, policy papers, statistics, and regulations; commercial sites for parts distribution; and engineering tools such as models and simulations," McGregor and Pacheco noted.