Automating a welding cell into a bay that had no room for it

Omar WatanabePredictive Maintenance Engineer · 19 yrs

The sequencing solution to the 180 A limit is the part of this worth copying, and the detail that makes it defensible is enforcing it in the control architecture rather than assuming it from the cycle. We did something similar on a two-cell resistance welding line and learned the hard way that a sequence which never coincides in the nominal cycle will coincide during recovery — an operator aborts, a weld is re-run, and suddenly two things that were never simultaneous are drawing peak together. Interlocking the peak-draw permissives so only one consumer can be in peak at a time makes the limit a hard constraint rather than a property of the happy path.

The other thing worth measuring rather than calculating is the actual peak draw, including the inrush on the positioner drive. Nameplate ratings and the sum of duty cycles will get you an estimate, but the current you have to design against is what a clamp meter and a data logger record over a full shift including faults and restarts. On our cell the measured peak was about 15 percent above the calculated figure, entirely from drive inrush during index, and on a supply capped as tightly as yours that margin decides whether the design works.

Samuel MehtaOperations Research Engineering · 15 yrs

Verified expertise in Operations Research Engineering

Simulating reach and cycle before specifying hardware sounds obvious and is genuinely rare, so it is worth saying why it paid here specifically. In an 11 by 8 m bay with maintenance access required on every side, the binding constraint is not robot reach — it is the swept envelope during the worst-case move, which is frequently a recovery or homing move rather than anything in the production cycle. Configurations that fit during production and collide during homing are a common and expensive discovery. If you simulated homing and fault-recovery paths as well as the cycle, that is the detail other people should copy.

On maintenance access, the number that matters is not clearance in plan but whether the largest consumable and the largest serviceable assembly can physically come out along a path that exists. Torch consumables, wire drums and the positioner bearing all have a removal route, and a layout that satisfies plan clearance while requiring the cell to be partly dismantled to change a bearing will be worked around by maintenance within a year. Worth checking against your final layout if you have not.

Dmitri YilmazSmart Infrastructure Engineer · 14 yrs

The works council point is the most transferable thing here and it matches my experience across several installations. The approval conversation is never really about the safety case — that is expected to be competent and is treated as table stakes. What decides it is whether the people in the affected area believe their own working conditions get better or worse, and fume extraction in a shared manual area is exactly the kind of concrete improvement that lands, because it is something they experience every shift and can verify themselves.

What has worked for us alongside that is bringing the operators into the load station ergonomics before the layout is frozen rather than presenting a finished design. Two or three sessions with the actual people on a mock-up, and the resulting changes are usually minor — part presentation height, which side the fixture loads from, where the reject bin sits. But the design arrives at the works council already carrying their input, which changes the meeting from a review into a confirmation. Retained headcount matters too, but it is only credible if the detail shows you thought about their day.