Lean Automation Selection and Its Limits

Why it matters

Lean automation selection asks which work is truly necessary before it asks which work can be given to a machine. Searching for a part, carrying it again because it was placed wrongly, or rechecking it because the acceptance condition is unclear does not gain value when done with a fast fixture. A method study must explain the purpose, sequence and safe execution of the work. The automation candidate is the defined work that remains after this review. Automating an unstable job can build daily corrections into software and hardware. Changing such an investment can be harder than correcting a simple layout. So the first deliverable is not a device list but a clear description of the accepted work.

Manual work, the time a machine is busy and the time a part spends at a station are different quantities. If K01 can do another job while a fixture is running, the times may partly overlap. If K01 must hold the part, the same machine time creates no free time. Automating one element also does not mean that loading, unloading, verification and routine intervention disappear. These new jobs are made visible first. Then a time axis is built together with the precedence relations and safe access. Simply subtracting the new machine time from the old manual time mixes the change in load with the change in cycle in one calculation.

Speeding up one station is not speeding up the whole line. If another station limits the output, a local capacity increase may only pile up stock in front of that constraint. A lower manual load can still be meaningful: strain may fall, another defined job may be taken over safely, or a margin against variability may appear. But each of these needs its own evidence. Converting freed seconds directly into wage savings is not correct. If the same person keeps working at the same pay, cash cost does not fall by itself. Capacity, manual load, safety and cash effect are shown in separate result rows.

An automation decision weighs the simplest workable option as well as the fastest one. Presenting a part in the right position may reduce the need for a complex handling device. The purpose of a trial is not to justify a device chosen in advance; it is to find the option that does the necessary work reliably. The acceptance condition, the authority to stop, the safe behavior on failure and the manual fallback method are defined together. If the method still changes often, stability comes first. The numerical evaluation also respects this limit: an attractive-looking return cannot make an unworkable or unsafe method acceptable.

The feasibility of a load distribution must also be examined independently of the assumption that a person has to keep doing the same work for the whole shift. Automating one portion does not remove the need for rest or the safe working limit. If the new arrangement asks the worker to intervene more often, the average manual work may look low while the attention load rises. So the pilot also observes intervention frequency, the overlap of several requests, and interruption and restart of work. If one person is proposed to move between different jobs, what matters is not only the sum of the times but when the jobs become ready. Even if the total time of two tasks that must finish at the same moment looks suitable, the task allocation may not be workable. The task transfer on which the economic decision rests must be shown with this detail. For the accepted method, alongside normal operation, the conditions for product changeover and restart after a failure are also described. Fast results obtained by bypassing a protective device or skipping a check are not suitable cycle samples. They are recorded but not made the basis of the new standard. A good automation decision requires showing that the suitable work is repeated reliably before choosing a low time.

Formula