The earliest applications of computers in manufacturing were for industrial control. Military technology developed during World War II vastly improved industrial controls technology. Especially important were new monitoring devices and electromechanical-servo systems created during the war. These technologies soon found their way into petroleum refineries and chemical plants. By 1950, IBM began selling small analog computers to the industrial world. Among the earliest devices used in discrete manufacturing were numerical control (NC) machines. John Parsons and others in the late 1940s pioneered the use of computer technology to control machine tools. Numbers recorded on punched cards guided the movement of NC machines in cutting parts.
Between 1950 and 1980, industrial firms introduced computer technology into virtually every major manufacturing activity. General Motors applied computer-aided design (CAD) techniques in 1960 to help create the tooling necessary to fabricate car body panels. Unfortunately, these industry efforts were done independently of each other, thereby creating “islands of automation.” Since the various departmental computers were not interconnected, interdepartmental coordination and communication were poor. Early attempts to rectify this problem centered on creating point-to-point interfaces between pairs of computers. An example would be transferring the data directly from a CAD workstation to the part programs that drive an NC machine. Joseph Harrington, Jr., and others, however, recognized that the problem was far broader and that only enterprise-wide integration would truly unify the highly fragmented computer activities in a manufacturing company. As a result, industrial firms now aspire to architectures and infrastructures that facilitate the integration of previously standalone systems.
The computational devices in a manufacturing plant are typically organized into three tiers:
At the lowest level, process-control sensors assess proximity, temperature, and other physical conditions and feed the information up the hierarchy. In a complementary but opposite direction, control information originating higher in the hierarchy issues directives. These instructions are sent to actuators at the process-control level. There they start and stop motors, open and close valves, and control other physical actions.
Managing the process-control level are two, important types of programmable machines, programmable logic controllers (PLC) and distributed control systems (DCS). These gather sensory data, interpret this data and send control signals back to actuators. PLCs are more prevalent in discrete operations such as assembly. DCSs are commonplace in process and batch industries, such as in glass plants. Both PLCs and DCSs have traditionally been special-purpose, digital devices designed expressly for manufacturing environments. Today, however, they are often based on the same generic, personal computer technologies found in office environments. Regardless of their pedigree, they are industrially hardened and optimized to handle considerable input/output traffic in a deterministic manner.
Next up the hierarchy are cell controllers. They coordinate the functions of multiple PLCs, DCSs, and other automated devices, such as robots, computer numerically controlled (CNC) machines, material handling systems, and machine vision systems. Cell controllers typically contain an operator interface. Through this interface a worker monitors cell activities and intervenes manually when necessary. An example of a cell controller would be a paint cell in an automotive paint shop. There the cell controller would direct the paint robot, check the paint job afterwards with a machine-vision system, and lastly signal that the cell was ready to paint another vehicle.
The highest level in the plant hierarchy are the area or plant-wide computers. These coordinate work among the lower-level cell controllers and serve as the interface to the external world. These area and plant-wide systems manage the overall flow of materials. They assign labor and machinery to specific production orders and monitor the production process.
From humble roots in the 1920s began the impressive march to the levels of automation now found in some leading edge plants today. Tying the three levels together remains a daunting task. This is despite major, past attempts at standardization. Among all areas of integration in the automotive industry, the plant remains among the most difficult to unify in a standard manner. The consequence in plants has been costly customized solutions as the rule, not the exception. Manufacturers able to crack this nut will enjoy substantial competitive advantage, not just in faster plant start ups, but also in significantly lower operating costs.