Direct answer: When a part is too large to lift comfortably, the alternative is to move the welding equipment to the part and leave the part on its stands. That changes what planning has to solve. In a fixed cell the part always arrives at the same place, so the working frame is stable; once the equipment moves, the frame is new at every stop, and parking position and datum recovery become the real work-position questions.
Who this is for: This guide is written for fabrication engineers and project managers welding large structural assemblies, where moving the part into a fixed cell costs more than the welding itself.
Scope: It covers how work positions, parking accuracy, datum recovery and site movement decide whether a mobile unit is the right answer, and what each judgement still owes in evidence. It does not cover welding procedure selection, structural design, or crane and lifting engineering.
When lifting costs more than welding
On large structural work the expensive step is often not the arc time. It is everything that happens before it: rigging the part, waiting for a crane, moving it down a bay, setting it down and levelling it, then doing the same in reverse. Each of those steps consumes people and floor time that never appears on a welding parameter sheet.
Mobile welding turns that around. In the reference footage a self-propelled unit carrying a welding arm parks beside a large weldment resting on low stands, welds the joint in place, then drives itself to the next part further down the hall. The parts never leave their stands.
This is not an argument against fixed cells, and mobile welding is not a replacement for them. It is a question about which side is harder to move. Where parts are small, repeatable and frequent, bringing them to a fixed station is clearly right; where they are large, awkward and few, bringing the equipment to them can remove the crane from the cycle entirely.
Moving the equipment moves the datum
A fixed cell has one quiet advantage that is easy to overlook: the part arrives in the same place every time, so the relationship between the robot and the joint is known before the program runs. That single fact is what allows a program to be reused without thought.
A mobile unit gives that up. Every time it parks, the relationship between the arm and the part is new, and it is new in a way that floor flatness, tyre position and stand height all contribute to. The seam is where it always was; what has moved is the frame the program describes it in.
According to ISO 10218-2:2025, the assessment covers the application and the cell rather than the robot in isolation, and with a mobile unit that boundary genuinely moves with the equipment. Each parking position is, in effect, a new installation of the same cell, and the checks that belong to an installation belong to each stop.
Parking position is the planning unit
This is why work-position planning for mobile welding is not route planning. The route between stops is the easy part. The planning unit is the stop itself: where the unit parks, how accurately it parks there, which seams that position covers, and how the seam is located again once it has stopped.
Locating the seam again can be done with fixtures and hard stops, with markings, or with sensing. The choice is a trade between preparation time and cycle time, and all of the options consume time at every stop. A plan that counts arc time and travel time but not relocation time will always look better than the cell behaves.
The number of stops follows from the same logic. Fewer stops mean fewer relocations but a longer reach from each one; more stops mean shorter reaches but more setup. Deciding that split for a given part is the substance of the work, and it is the part of a flexible welding for large structures proposal that deserves the most scrutiny.
Decision table: what the part and the site justify
The table maps what a part and a site can justify on their own against the evidence a supplier still has to produce.
| Evidence from the application | Selection it justifies | Evidence you still owe |
|---|---|---|
| Large parts, few per period, seams spread over the structure | A mobile unit with a station split, costed with relocation time inside the cycle | Trial at the least favourable stop, with relocation and seam location timed |
| Parts that must stay on their stands throughout fabrication | Work positions planned around the existing stand layout rather than around a cell | Floor and access survey, including stand height variation and clearance at each stop |
| Mixed sizes, some parts small enough to move | A split decision: fixed station for the movable family, mobile unit for the rest | Handling comparison for the movable family, so the split is a costed choice |
| Congested floor with shared routes and frequent foot traffic | Route and boundary planning before equipment selection | Site movement plan with the safety boundary defined per stop, not per cell |
Cables, routes and the people in between
A unit that drives around a shop takes its services with it. Power, gas and wire all follow the machine, and their routing becomes part of the plan: where they run at each stop, how they are managed during a move, and what they cross on the way. This is unglamorous and it is where these installations most often disappoint.
The other consequence is that the equipment now enters areas where people work. A fixed cell answers that question once with a fence. A mobile unit has to answer it at every stop, and the answer depends on how much room there is, who is working nearby, and what happens when someone approaches during a move.
According to ISO 12100:2010, risk reduction begins at the design stage rather than being added at the guarding stage, which for a mobile unit means the site movement plan is part of the equipment specification. In practice that plan needs the shop floor layout beside it, because aisle widths, temporary storage and stand positions decide what is achievable.
What the footage proves and what it does not
The reference footage proves a visible process. A self-propelled welding unit welds at one position, drives itself across the hall under its own power, and unfolds its arm again beside a different part; the parts stay on their stands throughout. Those are observable facts about the operation.
It proves nothing about parking accuracy, relocation time, deposition or the quality of the resulting joint. According to ISO 3834-2:2021, comprehensive quality requirements for fusion welding are established through documented arrangements rather than inferred from an observed operation, and that applies to a mobile installation as much as to a fixed one.
Naming that boundary is what keeps a comparison honest. What the footage settles is that the concept works and what it looks like; what remains is a trial at your least favourable stop, with relocation and seam location measured rather than assumed. EVST treats those two things as separate deliverables for exactly that reason.
What to put in the enquiry
State the part: dimensions, mass, how it is supported, and where the seams sit on it. Add the number of parts per period, because a mobile unit that relocates constantly for a handful of parts is a different proposition from one that works through a steady queue. Comparable context is worth including as well, such as a large workpiece pose and reach test or existing steel structure welding automation on adjacent products.
Then state the site: floor condition, aisle widths, available parking positions, stand heights and the routes the unit would use. Finally state what the supplier has to prove: a trial at the least favourable stop with relocation and seam location timed, a services plan, and a movement plan with the safety boundary defined per stop. With those in hand a mobile unit and a fixed station can be compared on the same evidence rather than on preference.
Frequently asked questions
Is a mobile unit slower than a fixed cell?
Per seam it usually is, because relocation and seam location are added at every stop. The comparison that matters is total time including handling. Where a fixed cell needs a crane, rigging and setting down for each part, a mobile unit can still finish first.
How accurately does the unit have to park?
Accurately enough that the seam can be located again within the time you have allowed for it. That is a system answer rather than a specification: a coarse parking position with fast sensing can outperform a precise one with a slow location routine.
Can one unit serve parts of very different sizes?
Often yes, but the station split changes with the part, and so does the reach from each stop. Treat each part family as its own work-position plan rather than assuming one plan transfers, and check the least favourable stop of each.
Does the footage show how long relocation takes?
No. It shows the unit welding, driving itself to the next part and unfolding again. Relocation and seam-location times depend on your floor, your stands and the location method chosen, and they have to be measured on site rather than read from a video.
Project inputs for an application review
Send the following and the work-position plan can be checked against your floor instead of a generic layout:
- part dimensions and mass, how it is supported, and the seam map
- the number of parts per period and the sequence they arrive in
- floor condition, aisle widths and the parking positions available
- the services available at each position, including power, gas and wire supply
- who works nearby, and what has to happen when someone approaches during a move
If you have large structures that are awkward to lift, send the part dimensions and seam map, the production sequence, the floor and access survey, the services available at each position, and the movement pattern of people around the area. The station split, parking positions and relocation time can then be planned on real conditions. Related reading: steel structure welding automation.