Production Automation

“…if every instrument could accomplish its own work, obeying or anticipating the will of others… if the shuttle weaved and the pick touched the lyre without a hand to guide them, chief workmen would not need servants, nor masters slaves.”  Aristotle 

A working description of production automation

A brief history of automation

A working description of production automation

For our purposes, production automation may be defined as the specialized combining of several, sometimes-disparate disciplines or technologies in a machine to make performing a production task (or portions of a task), or process, autonomous to some degree.  The degree of autonomy may be as small, or as great, as is needed to achieve a desired end. 

For example, though a drill jig is not commonly thought of as automation, it does indeed make precisely locating a work piece under a drill “automatic” for an unskilled workman. Some of the skill needed to perform the drilling task is embodied in the jig. A well-designed and fabricated drill jig increases quality, increases production speed, and reduces the qualifications needed by a worker. The jig is not automated, the operator must provide the motions, but some of the drilling task is accomplished automatically, that is, without effort or thought by the operator. Though the automation is not the “self guiding” or “self actualizing” kind, the desired end result is achieved.

Consider a common milling machine that controls the instantaneous degrees of freedom of a part being made. It does so while the operator controls the path of the part by turning the positioning screws. From the operator’s point of view, dumb metal is acting autonomously to control some aspects of the task – that of constraining several degrees-of-freedom. The autonomy is only perceived, it is not real, but the effect is the same as if it were real. The finished part is made with a speed and quality not possible for an artisan equipped only with a file, no matter what level of skill and experience is available. Thus the machine is not automatic, but some aspects of the machining task are handled automatically. 

At the other end of the automation spectrum we have complete machine autonomy. Consider an unmanned missile that is launched from a silo in the heart of a large country. It may have to fly thousands of miles to a target in an enemy country. It must accurately reach its target and detonate its warhead in a manner calculated to cause the desired degree of damage. Such a missile is an autonomous machine from the time it receives its launch command until it explodes its payload. The missile is not self-aware; it does not think, but it is a self-actuating machine that is able to achieve a built-in goal without outside guidance. 

Almost all practical industrial automation will embody a degree of automation, or autonomy, that falls well within the range defined by the above rather extreme examples.  For example, consider a bottle filling line. An operator must dump the bottles into a bin for sorting and feeding. He must dump the caps into a bin for sorting and feeding. He must put the liquid product in the filler’s tank. He must put the preprinted labels in the labeler. And he must put a stack of collapsed cartons in the cartoner’s magazine. 

In such a line, several well-engineered, semi-autonomous machines (filler, labeler, etc.) are interfaced sequentially (mechanically and electrically) to form a larger, semi-autonomous machine. When all is well, the line runs autonomously for a while. An operator/observer is available to act when events occur which are outside the autonomy of the machine.  

The above description is an example of a common approach: that of building an automated line from available, proven special-function machines like fillers and labelers. There is another aspect of industrial automation that must be mentioned to give a fair representation of the field. Suppose a company invents a new product, the widget. The fabrication and assembly of a widget requires some tasks that are not specifically common to industry. Several small spring-loaded parts must be sequentially attached within a small cavity in the widget. Because the widget is a new invention, there are no third generation widget assembly machines commercially available. Either an existing machine will have to be re-engineered (usually not practical) or a custom machine will have to be designed and fabricated. A custom machine may have to incorporate a new and inventive way of doing something if the need is novel enough. 

This is a good point to introduce the concepts of flexible automation and fixed automation. If a new machine is designed to accurately assemble the new widgets at high speed, it is an example of fixed automation. It does its intended task very well, but is probably unsuited for any other duty. Its functionality is fixed. Because the machine is tailored to widget assembly specifically, each mechanism, each actuator, each component material, etc., and their synergistic combination, can be optimized. Therefore, widget quality, production speed, system fault diagnostics, personnel safety, and system reliability can be optimized. 

Commonly available industrial robots exemplify flexible automation. These are often relatively small, single-arm devices that are well suited for such applications as small parts pick and place and spot welding, for examples. They are a form of flexible automation because they can be taken out of one application, reprogrammed, and used for a different application. It may be practical to use several such robots (each with custom programming for widget assembly) to perform some of the widget assembly operations. When it is practical to do so, there are several potential benefits and several potential penalties.  

If flexible automation is practical for widget assembly (considering cost, size, programmability, interfacability, safety, production speed, etc.) some potential (but not automatic) benefits are:

          Lower initial costs compared with custom design

          Re-usability of the machines for other applications

          Economy of scale (purchasing several identical machines)

          Faster implementation compared with design and fabrication

 Even in the most industrialized nation in the world there are far fewer industrial robots in use, than were predicted just a few years ago. There are reasons for this. Such robots can solve some problems very well. However, trying to fit a robot to some tasks can be akin to using a Swiss army knife to make a door for a house. It can be done, but it takes significant skill and time, and it is obviously not the best way to do the job. Sometimes clothing is advertised as “one size fits all”. What it means is: one size can be worn by all, but it will “fit” none well. General-purpose robots do not fit all sizes well. 

Robots come in many sizes, but they are usually designed for general usage. They are not designed specifically for making widgets; their designers have never seen a widget. So the best available size and configuration will have to serve. In most cases, production speeds achieved with robots will be much less than with fixed automation. Specialized knowledge is required for the implementation of some robots. This can result in unexpected expense and delay. 

The preceding description is of general-purpose robots. Often, a custom robot will be designed for a specific application. For example, a simple gantry type robot may be designed for a carton loading application. Such a simplified robot may still be able to be used for other similar applications, and thus be considered flexible automation.

As with many types of machines there is a range of cost and complexity associated with robots. A specialty “idiot” robot can often be economically designed and implemented for a given application. It is an idiot, because it only has enough designed in functionality to do its intended job (actually, it is a genius for its own job, and an idiot for all others). The kinematics and software are suited for the application at hand. The expense of unneeded functionality can be saved. Such robots can be very effective for a specific application, and can cost less than more general-purpose models. For certain applications they also may be significantly faster acting. 

With the current state-of-the-art most anything that needs to be made fully automatically can be. However, sometimes the cost is prohibitive for full automation. Read further to gain a better picture of the benefits and costs of automating a task or procedure, and of factors that can minimize the costs while achieving the needed functionality.

A brief history of automation

For most of recorded history, production automation was neither practical (the technology and materials were not available), nor necessary; human labor was cheap and abundant. As an old joke observes: a human being is the cheapest form of automation, and the only kind that can be produced with unskilled labor. It was not so long ago that many were willing to labor for their daily bread.  

Leonardo Da Vinci bent his genius to automating a few tasks, such as pumping water [Illus.], grinding mirrors [Illus.], and making files. Though undoubtedly a genius of high order, most of his designs were speculations, and, as he drew them, impractical. They are, however, noteworthy for their uniqueness in a time when machinery and visionary thinking of production were very uncommon. 

Leonardo would have been recognized as a great engineer had he lived in a time when gears, bearings, steel, and motors and engines were available.  When Leonardo conceived of a machine, he had to conceive of a way to accomplish all aspects of the machine’s workings. He was greatly hampered by limited materials and primitive fabrication tools. Considering his talents and aspirations, he was born out of time. 

One of the earliest known examples of practical automation is due to an unnamed genius who harnessed wind to grind grain. The same technology was soon adapted to pumping water. It is reported that windmills were used in England and Holland from the 12^th century. The self-acting pumping of water from the Low Countries enabled Holland to reclaim valuable land from the sea.  

A critical mass of needs, talents, materials, workmen, and markets developed in the 18^th century that brought about the industrial revolution and the beginning of practical automation. In 1745, Edmund Lee, a blacksmith, invented the fantail [Illus.], a device that automatically kept a windmill’s blades pointing into the wind. This improvement greatly increased the amount of power available and removed a need for human intervention. Although this invention is largely overlooked, it is a significant step forward in the field of self-correcting or self-guiding machines. 

Wind is an uncertain and varying power source. The grinding stones of early windmills turned at a rate determined by the wind velocity. As the stone speed increased, the dynamics of the grain flowing between the stones caused them to move further apart, thereby degrading the quality of the flour. A means was needed to automatically regulate the speed of the stones while still drawing power from the wind. 

In 1787 Thomas Mead patented a means of regulating the speed of windmills. He used a centrifugal pendulum to tie the effective area of the windmill sails to the speed of the turning stones. Thus, if the stones increased in speed, the sail area was decreased, slowing the blades; if the stones decreased in speed, the sail area was increased, speeding up the blades. This invention was one of the first really practical, and valuable, applications of the principle of negative feedback for speed control. Much of modern machine control, and therefore automation, is based upon the principle of negative feedback. 

James Watts was a key player in the industrial revolution. He greatly refined the steam engine in 1769, increasing its power, efficiency, and scope of use. His steam engine was the prime mover that drove 18^th and 19^th century industry. For a prime mover to be serviceable, its output speed must be controllable. In 1788, Watt invented the justly famous centrifugal flyball governor [Illus.] (it was based in part on the concept of Mead’s patent). Watt’s governor coupled with his engine make him one of the most influential inventors of all time. Because of Watt, the world could turn coal or wood into precisely controlled power and speed. 

Some historians consider Oliver Evans to be one of the founding fathers of American industry. He certainly has some claim to the title of  “Father of American Automation”. In 1790 he received the third U.S. patent for an automated flourmill. [Illus.]  Mills built to his design could accomplish with two men what prior designs required eight men to accomplish. Additionally, there was far less waste for an Evan’s mill, and its product was of obvious superior quality. 

By automating tasks and joining them by feeding screws and conveyors he automated an industrial process. His flour factory functioned as a machine, with material going in, and a valuable product coming out. By applying the principles of available technology, he reduced the production costs; he reduced the uncertainties associated with employees; he increased the production rate; and he improved the quality. An Evan’s mill owner could produce a better product for a better price, not a bad circumstance for a businessman, then or now. 

Oliver Evan’s relative obscurity is strange considering the value of some of his inventions. He invented a high-pressure steam engine that made Mississippi River commerce possible. That invention alone made him an extremely valuable citizen of his country. Unfortunately, most of the engines of his design that plied the great rivers were pirated. He received no compensation for them. It is strange that inventors of lesser merit achieved greater, and more enduring, fame and recognition. 

Thomas Mead’s invention was a valuable instance of the application of negative feedback to achieve control. In the first few years of the 19^th century, Joseph Jacquard of France developed the valuable and influential Jacquard loom. [Illus.] [Illus.] His loom used punched cards to automate the weaving of complex patterned textiles, including carpets. The punched cards directed the motions of the machine, eliminating the need for skilled workers to produce quality textiles. This means of machine control is based on positive feed forward, which is complementary to the principle of negative feedback mentioned above.  

Much later Herman Hollerith, in the U.S.A., used the information storing capability of punched cards to great advantage. His company – later to be known as IBM – became a leading force in the computer revolution. Jacquard’s use of punched cards for machine control is very strongly analogous to software control of machines today. 

The above examples are illustrative of doing more with less, and doing it better, by building autonomy into machines. Obviously, many other individuals and inventions of great merit and import could be cited. For examples: England’s Portsmouth Dockyard pulley factory created by Brunel and Maudsley, the development of modern machine tools by Maudsley and Naysmith and their contemporaries, and Whitney’s practical interchangeability of parts and custom production machines in America. There are others too numerous to list in a brief history. 

Practical industrial automation really began in the late 18^th century. It developed rapidly thereafter. The two world wars spurred advances in science and technology that were quickly adapted to industrial needs. Engineers of our era have no greater intelligence than those pioneering engineers listed above, but they do have a much better collection of tools (physical and intellectual) to work with. 

Photo eyes, Hall effect sensors, servomotors, engineered materials, pneumatic components, hydraulic components, bearings, and a host of well engineered components, prime movers, and sub systems too numerous to mention are now readily available.  Low cost, high performance computers coupled with powerful high level computer languages like C and C++, and simple and powerful control methodologies like PID and Fuzzy Logic permit a degree of automatic control previously not possible. Today’s machines can have eyes, ears, hands, feet, and brains suitable for almost any given application.   

BACK TO TOP

Practical and cost-effective automation

As mentioned above automation can be of varying degrees and scope. A sewing machine automates a task: that of stitching two or more layers together. This level of automation brought great benefits to society and provided livelihood for many individuals. The first practical sewing machine was a great inventive step forward. It greatly magnified the productivity and skill of the operator and resulted in a much more uniform quality in the product. 

A greater degree of automation is possible today than was possible when Elias Howe first offered a practical sewing machine. Today it is possible and practical to automate dress making: to scan the customer for a perfect fit, to cut the material to a proven pattern, to position the pieces and to stitch and even embroider them. 

However, although it is often possible (and occasionally needful) to completely automate some processes, sometimes a lesser degree of automation is more practical and cost effective. Piecewise automation may be defined as the automation of only a part (or parts) of a whole task or process. Thus, piecewise automation might be applied to the most dangerous, or most quality sensitive, or most labor intensive parts of a complex assembly operation. Given available resources, piecewise automation may be the only realistic approach to solving a given production problem. 

The true value of automation for a given application depends on the problems it solves. Not all problems are equal. For a medical application, quality trumps labor costs. For an explosives production line, safety trumps production speed. For a rubber band packaging line, labor costs may be the sole criteria. For some applications, removing some of the labor may make a lot more sense than removing all of the labor. 

The following are common reasons for justifying an automation project:

·       Increased production speed

·       Increased product quality and uniformity

·       Increased sanitation

·       Increased employee safety: some jobs are too dangerous

·       Elimination of repetitive motion tasks: carpel tunnel issues, etc.

·       Decreased product cost

·       Fewer human resources problems

·       Elimination of distasteful tasks: boring, filthy, smelly, etc. 

When you decide to have a home built, you can build your dream house or you can build to a budget.  These are usually not compatible approaches. The need for automation can be driven by various factors. If the needs are truly and accurately known project specifications can be judiciously determined, and satisfactory solutions derived. If they are not, it is easy to waste resources, and even to solve the wrong problems. If enough automation is just right, too much is not better. Too much automation often leads to project failure. True needs must be balanced with available resources and with justifiable pay back expectations.