The Human Touch

Whether it’s designers cleverly developing products for robotic assembly or skilled workers performing intricate operations alongside sophisticated technology, the human factor is key to successful factory automation.

For automation to be effective, products should be designed with automation in mind. While that statement may seem rather obvious, in the past, components were designed simply to work. That is, engineers concentrated on making the component do the job it was supposed to do; how to assemble it was an afterthought.

This was mostly a matter of making sure that assembly was possible, then leaving it up to human assemblers to figure out the fastest way to do it. It is more recently that design for manufacturability (DFM) concepts have come into vogue, charging designers to think about how their creations are to be manufactured right from the start.

Segmented core stator assembly

Another key consideration in factory automation is evaluating which processes to automate and which to leave in the hands of the human operator.

Anticipating automation

Mitsubishi Electric’s Himeji Works in Japan is a good example of designing with automation in mind. The factory assembles a wide range of automotive alternators and other products via an almost completely automated production system; only 11 employees are on hand to assemble thousands of alternators per day.

Himeji Works practices a variation of DFM that might be called “design for robotic assembly”. Constraints for automated assemblers are quite different and more restrictive than those for humans. For instance, it’s quite natural for a human to pick up a subassembly and turn it over to reach the bit that needs to be worked on. But take away the human, picking the subassembly up off a conveyor, then turning it over requires quite a complex machine. And doing it more than once on the same unit involves even more complexity.

While coil winding on free-standing forms was one of the first manufacturing operations to be automated at the dawn of the electronics industry, adding copper coils to the magnetic poles of an alternator stator has always been an awkward process, even for human assemblers. The difficulty comes from the assembler having to stick his hands (or robotic grippers) into the ring-shaped laminated-steel core to wrap copper wire over each pole.

At Himeji Works, Mitsubishi engineers realized that winding coils would be much easier if the stator were inside out. And they managed to figure out how to turn an alternator stator inside out while winding the coils, then turn it back for final assembly.

Instead of forming the core in one piece, they form it in sections that hinge together. A robotic handler presents each of the segments in turn to a coil winding machine, which adds the copper wire as if it were on a free-standing form. After winding one coil, the handler indexes to present the next segment in the same place for the winding machine. After all coils have been wound, the stator goes to a machine that bends it around a form into a circle and welds the hinged joints to keep the stator’s shape.

Man or robot?

The broad guideline is that humans should do only those jobs that robotic equipment cannot, or that are too complex to program effectively. For example, at Mitsubishi’s Himeji Works, automated winding machines form the stator windings from copper wire, but it takes exquisite eye-hand coordination to slip the wire ends into the connectors. Currentgeneration robots cannot do it, so people must.

Of course, when anything goes wrong, robots have to call on people to help. Automated machinery is extremely capable when everything they have to deal with stays within tight tolerances. But when parameters get too far out of bounds, robots do not have the flexibility to analyze the situation and make a plan for bringing it under control. Only humans can do that.

Also, the cost of designing, installing, and programming automated production equipment typically is very high. To justify this expense, the resulting automated system has to produce a huge number of units at reduced cost with little or no intervention. Human labor is largely a variable cost while automation is something of a fixed cost, which means once you make the investment, the equipment can crank out as many units as possible for little or no additional outlay.

To compare the two, you must first amortize the automation costs across the number of units that the machine can reasonably be expected to produce during its service life. That is, estimate the number of units it’s likely to produce before needing major overhaul or replacement, then divide that into the machine’s cost of ownership.

That done, you can compare apples to apples, because both costs are on a per-unit basis. When the task complexity is high, the automation cost can become extreme. In that case, the calculation will tell you that it’s less expensive to have humans to do the job.

In other cases, such as when products are highly customized, you’ll never sell enough units to pay for the automation. Again, the answer is to hire humans – perhaps even paying premium wages for expertise – to do the job.