Production decisions
Automating a manual operation: what to look at before commissioning a machine
The usual mistake is not automating too late. It is automating the wrong action.
When a manual operation becomes a bottleneck, the conversation always goes to the same place: “we should automate this”.
And it is almost always framed the same way: take what the person does and give it to a machine. Same movement, same sequence, but without hands.
That approach is what produces expensive machines nobody uses.
Because the right question is not “how do I do this without people?”. It is “why is this operation causing problems?”. Often the answer is not speed or labour cost: it is that the method itself is damaging the product.
An example where the problem was not what it seemed
Stacks of sheets needed to be squared and joined before lamination and die-cutting. They were joined by hand with heated soldering tools, point by point.
From the outside, it looked like a productivity issue: a slow, repetitive manual operation.
Up close, the problem was different. Two problems, in fact:
The tips pierced the layers and left an irregular, raised exit point. That local extra thickness could damage the plates in subsequent operations.
The heat produced carbon residue. Particles remained trapped in the material and appeared after pressing as black specks. Rejection.
In other words: the manual operation was not just slow. It was manufacturing downstream defects.
If you automate the thermal spot-welding — same action, no hands — you take the raised areas and carbon residue with you. You have spent money producing the same rejects faster.
Change the action, not just who performs it
The solution was to change the joining process: ultrasonic welding instead of thermal spot-welding. Without heat, there is no carbon residue. With a properly formed weld, there is no raised area.
And around that, a semi-automatic machine that squares, holds and welds in a single cycle: interchangeable stops for working from either side, clamps that secure the layers before contact, and three independent points with timing set on screen.
Notice the order of the decisions. First, understand what was failing. Then choose the process. The machine came last, not first.
What almost nobody asks for, but determines whether the machine is used
This is what separates a machine that is used from one that ends up under a cover.
Hands free. A foot pedal triggers the cycle so both hands remain available to position the stack. There is also a push button for jobs that do not need that freedom. It looks like a detail; it is the difference between a comfortable and an awkward cycle repeated a thousand times a day.
What happens before and after the machine. Two material stands were added: sheets are picked up from one side and the welded stack is placed on the other. They can be turned and tilted — tilting helps square the sheets by gravity — and one also adjusts in height to bring the work to each person. On wheels, so material can be moved.
None of that welds. All of it determines whether the workstation works.
Put a check within the operator's reach. An optional air jet helps reveal an insufficient joint visually. It is not automatic quality inspection and is not presented as such: it gives the person working a quick way to look.
Prevent the error, not just work quickly
A later development added something worth explaining separately: a reader plate recessed into the table checks the material's chip before allowing the cycle.
If the chip does not match the selected job, the system blocks the downward movement and gives a warning. An independent screen next to the main one displays the result.
That goes beyond automation: it prevents an error that was previously possible. And it is exactly the sort of improvement that only occurs to you once the machine has been running for a while and you have seen what goes wrong in real life.
What is gained, stated honestly
In this case, the operating experience gathered indicates that the workflow went from needing approximately six or seven people to two. That is an observation from use, not a measured productivity study, and it should be presented that way.
It also indicates that carbon residue stopped appearing as a cause of waste, and breakages attributed to the raised welds from the previous process stopped occurring.
I prefer to state it that way, with its limits, rather than give a neat percentage that sounds good but cannot be defended. If someone promises an exact improvement before seeing your process, be sceptical.
What is deliberately not automated
A semi-automatic machine is not a half-finished automatic one. It is a decision.
Here the person continues positioning the stack because the material arrives with variations that automatic feeding would need to handle with much more machinery, much greater cost and much less flexibility to change jobs.
Automating 80% well usually works out better than automating 100% poorly. It also leaves room to keep improving: the first version of this machine dates from 2022, and later interventions have incorporated material validation, interchangeable consumables and a new clamp configuration with extraction for work involving materials that release gases.
Real use guides the improvements. Not the original plan.
Questions before asking for a quotation
- What exactly is failing? Speed, cost, repeatability — or is the method producing defects?
- Is the current action the right one? If the process causes damage, automating it multiplies the damage.
- What happens before and after? If the machine is fast but the workstation awkward, you have gained nothing.
- Which error do you want to make impossible? Preventing a failure is often worth more than saving seconds.
- Which part should remain manual? Deliberately, and with a clear reason.
At 7Axis Industrial we develop machines and purpose-built equipment when no off-the-shelf solution fits. Mechanics, manufacturing, electrical work, control and commissioning under one responsibility.
View the full ALZA+ case → Automation and purpose-built equipment →