7 The Numerical Control era.

Beretta acquired its first numerically controlled (NC) machines in 1976. These machines functioned automatically, performing operations and changing tools according to numerically coded instructions. Although this technology had begun to spread through Italy at the beginning of the decade, its presence was isolated. Its primary users were companies that manufactured small quantities of products of high value. With the introduction of microprocessors, controllers went down in cost and up in reliability, making NC technology viable for large scale use. Beretta introduced these systems into the high volume production (200 to 400 pieces per day) of small- to medium-size products.

Beretta regarded the automation of tool changing as the single most significant benefit of NC technology. Automated tool changing meant that what had formerly required a transfer line could now be accomplished with a single machine. NC machines (see Figure 7.1), which combined the versatility of general-purpose machines and the productivity of special-purpose machines, also overcame limitations imposed on particular components by the specialization of transfer lines. But they were expensive. At the time, the best Beretta was able to obtain was a four-year payback, and some had an eight-year payback.

[FIGURE 7.1 OMITTED]

"As you can see," said company president Ugo Beretta, "it was not what you would call a brilliant investment. But we had to do it sometime. We could have waited, but we could not turn back the clock. It was a very new technology, with electronics and computers, and we had to understand it. Instead of waiting, we decided we would go ahead and buy a machine tool company and learn the new technology. So we bought MIVAL, a small machine tool company with expertise in this field." (1) The effects of numerical control at Beretta are summarized in Table 7.1.

Numerical control had evolved out of a program funded by the United States Air Force in the late 1940s for making complex shapes. Although the first commercial products were offered a decade later, there was no significant penetration of NC systems until controllers became more economical and reliable in the 1970s. Although self-directed machines--automation--went back many decades, a critical distinction between NC and earlier automation was that the sequence of tasks could be easily altered or replaced by loading a new program.

The work cycle of NC machines--the set of motions that determines the selection of tools, their proper positioning in three dimensions relative to a workpiece, feeding of workpieces, flow of coolant, and so forth--was recorded as a series of codes initially on punch tapes, then on magnetic tapes. This information, called a part program, was passed to the "programmable controller," a crude special-purpose computer that processed the information and issued signals to the various motors on the machine to position the machine axes accurately and precisely, and cut the workpiece. The motions a machine tool must go through to produce a part must be described in detail, mathematically. This reduces the entire process of producing a part, including the skill of the machinist, to a formal, abstract expression, which, when coded and translated by a microprocessor, activates a machine's controls. Every machine movement, however slight, has to be formally, explicitly, and precisely articulated. With such programmable automation, a switch in products no longer entails physical setup changes to retool or readjust the configuration of the machines, only a switch in programs. Thus, NC technology combines the versatility of general-purpose machines with the precision and control of special purpose, or self-acting, machines. [31]

"In the past," observed The American Machinist in 1973,

humans were both translators and transmitters of information:

the operator was the ultimate interface between design intent, as

incorporated in a drawing or instruction, and machine function.

The human used mental and physical abilities to control machines.

Today, computers are increasingly becoming the translators and

transmitters of information, and numerical control is perhaps most

representative of the kind of control that plugs into the greater

stream with a minimum of human intervention. Historically,

numerical control certainly has been the most significant development

of the electronic revolution as it affects manufacturing.

Quoted in [29, p 221]

NC technology, after two decades of disappointment, came into its own with the advent of microprocessors. Microprocessor technology made controllers at once extremely powerful and relatively inexpensive and its greater computer power made possible sophisticated, yet flexible and "user friendly," operator interfaces. (2) It also made possible advanced control techniques including allowing NC machines to record utilization and cutting tool life, reduce set-up efforts and time, compensate for errors, inspect surfaces and make automatic adjustments, allow operators to modify their programming on the shop floor, record events of the last minute or two prior to a failure, and perform self-diagnosis. Coupled with greater sophistication in machine tool design, numerical control using microprocessors made possible the development of standalone machining and turning centers capable of shift-long, untended operation.

The early problems of NC technology were partially due to limited formal knowledge of the machining process. A lot of the knowledge possessed by skilled machinists, such as when and how to make "on the fly" adjustments, was tacit or otherwise not accessible to programmers. (3) This limited understanding of contingencies and variations in factors such as machinability, tool wear, and part material properties significantly constrained early implementations of NC technology. But with effort, over time more of the tacit knowledge implicit in operator skills became precise, explicit knowledge that was used to develop procedures capable of avoiding or dealing with a variety of contingencies.

7.1. NC Technology at Beretta--From Synchronous to Cellular

What happened to the organization of work in the Beretta plant after the installation of NC machines is interesting. In the transfer-line the average cycle time for a product was two minutes. Half of this was attributable to the machine, the other half to the operator. With the automation of tool changing, a variety of operations could be done by a single machine, but the overall cycle time increased. The cycle time required by an equivalent NC machine to perform the operations that previously required three machines would be 3.6 minutes, only .6 of which would be operator time. Thus, one NC machine replaced three machines, but took almost twice as long to produce a single part.

One can see that an operator of this NC machine would be idle 85% of the time (3 minutes out of 3.6 minutes). By allocating two machines to each operator, he would be busy 1.2 minutes while the cycle time would still be 3.6 minutes. Thus his idle time could be reduced to 66%. The greater the machine component of the cycle time, the larger the cluster of machines it makes sense to put around the operator. This leads to a cellular rather than synchronous plant layout.

In a synchronous line with two-minute cycle times, an operator performed a fixed, unchangeable routine. The nature of the work was determined by "hard automation," the jigs, fixtures, and cams that governed the performance of the operation. With hard automation, considerable effort was expended to get the jigs and fixtures right the first time. "Quality" was front-end loaded in the hardware design and quality control was a process of monitoring and tending the machines and tools.

The scope of activity at any given workstation was very small and the machine established the pace of work. The principal intellectual activity on the line consisted in monitoring machine performance and diagnosing problems when they occurred. Because a problem at any one station on a synchronous line could stop all subsequent operations, thus exacting a high cost in productivity, a large and centralized set of resources was allocated to problem solving. At Beretta this allocation was seen in the growth of the Quality Control department.

A cellular plant layout significantly increases the scope of activities for which an operator is responsible. The twelve operator stations in the barrel line layout shown in Figure 7.2 are responsible for one hundred and sixty-eight operations, an average of fourteen operations each. This compares with an average of three operations per person on a typical indexing machine in a transfer line used to manufacture, for example, the Garand rifle. Thus, we have a five-fold increase in the scope of activities.

[FIGURE 7.2 OMITTED]

We find, too, that the nature of the work changes. An NC operator works not with physical objects, but with information. The object of attention and medium of work is a computer program. Whereas the operator on a synchronous line was interested in observing the behavior of a process, the operator in a manufacturing cell composed of NC machines is interested in observing the behavior of a procedure.

7.2. Softening "Hard" Automation