Direct answer: In prefabrication automation the bottleneck is rarely one machine. Rolling, forming and welding each automate well on their own, so adding up machine capacities overstates what a shop actually produces. Output is governed by the handover: how parts move between operations, where they wait, whether batch rhythms match, and how many machines one changeover stops. Cost the handover before the machines, and automate the constraining operation first.
Who this is for: Written for fabricators automating ductwork, shells, cones and similar prefabricated assemblies across several separate operations.
Scope: This EVST guide covers why single-machine capacity overstates shop output, which handover questions decide real throughput, and what to prepare before comparing quotations. It does not size any machine, quote throughput, or design a layout.
Three operations, three machines
The reference footage shows three separate scenes: a plate rolled into a cylindrical shell on one machine, a cone rolled on a second, and a robot welding a branch joint onto tube.
These are three different machines working on three different parts. They are shown together because prefabricated products typically require all three kinds of operation, not because they form one continuous line.
Reading them as a line is the mistake this article is about, because prefabrication automation is bought machine by machine and judged shop by shop.
Single-machine automation is the easy part
Taken separately, each of these operations has well established automation. Rolling has dedicated machines, forming has dedicated machines, and welding can be robotised.
That maturity is precisely what makes capacity addition tempting: three machine ratings, one sum, an apparent shop output.
The sum assumes parts arrive at each machine exactly when it is free, which is the assumption the shop floor does not honour. According to ISO 10218-2:2025, the application is what gets assessed rather than the individual machine, and prefabrication automation behaves the same way when output is being estimated.
The handover questions
How does a part get from one operation to the next: crane, trolley, conveyor, or manual carry, and how long does that take relative to the machine cycle?
Where does a part wait, and how much space does waiting need? Buffers that do not exist become floor space that does, or machines that idle.
Do the batch rhythms match? A rolling machine that produces one shell at a time feeding a welding station that sets up several parts per fixture is a mismatch that shows up as waiting, not as a capacity shortfall.
How many machines does a changeover stop simultaneously, and how often does the product mix force one?
Sequencing prefabrication automation
Where budget forces a phased approach, the operation to automate first is usually the one that constrains the others, not the one with the most attractive machine.
Automating a fast operation upstream of a slow manual one typically moves the queue rather than shortening it.
According to ISO 3834-1:2021, criteria are set out for selecting an appropriate level of quality requirements for fusion welding, and where welding is one operation among several in prefabrication automation, deciding that level early keeps the welding step from being specified in isolation. According to ISO 12100:2010, the limits of the machinery are determined before protective measures, which in a mixed shop means the handover has to be described before any single machine is specified. Acceptance of the result should be written against the linked flow, not against one machine in isolation.
What the footage proves and what it does not
It proves three visible operations: a plate rolled into a shell, a cone rolled on a different machine, and a robot completing a branch joint weld on tube, with the finished joint visible afterwards.
It does not prove that these operations sit in one facility, that they are connected, or anything about throughput, utilisation or cycle time. The narration and this article describe them as separate operations on separate machines because that is what they are.
Any integrated output figure has to be calculated from your own product mix and volumes.
A short decision table
One dominant product with stable volume: linking operations physically is usually worth costing.
High mix with frequent changeover: buffering and changeover time often deliver more than additional machine speed.
One operation clearly constraining the rest: automate that one first and re-measure before committing to the others.
| Situation | Priority | Why |
|---|---|---|
| One dominant product, stable volume | Physical linking | Transfer cost repays across a long run |
| High mix, frequent changeover | Buffering and changeover time | Machine speed is not the constraint |
| One operation constrains the rest | That operation first | Automating elsewhere moves the queue |
What to prepare before asking for a quotation
Product structure and size range: shell diameters and thicknesses, cone sizes, joint forms on tube. Annual volumes and typical batch sizes per product type.
Existing equipment list and shop layout, including crane coverage and available floor area for buffering.
Changeover frequency, staffing per shift, and which operations currently run overtime.
Frequently asked questions
Why not just add up machine capacities?
Because that assumes parts arrive exactly when each machine is free. Transfer time, buffering and changeover break the assumption, usually by a wide margin.
Which operation should be automated first?
Normally the one constraining the others. Automating a fast operation upstream of a slow one moves the queue rather than removing it.
Do the operations have to be physically linked?
Not necessarily. For high mix work, buffering and short changeovers often deliver more than a rigid link.
How is welding quality handled across a mixed line?
By deciding the quality requirement level early, so the welding step is specified alongside the others rather than in isolation.
Project inputs for an application review
If you are planning prefabrication automation, the product mix and the handover picture matter more than any single machine specification. EVST costs the handover alongside the machines.
- product structure, size range and joint forms
- annual volumes and typical batch sizes
- existing equipment list and shop layout including crane coverage
- changeover frequency, staffing and current overtime
Send those and EVST will work through sequencing and handover against your actual product mix. Related reading: automatic welding systems, positioners for long cylindrical parts.