Large Plate Welding: Travel First, Arm Second
Direct answer: On a large flat weldment the binding constraint is reach, not speed. The part stays on a floor platform, a gantry spans it, and the robots hang inverted under the crossbeam. Inverting frees the floor beside the part and routes services overhead. Coverage then comes from gantry stroke rather than arm reach, which reverses the usual sizing order. Two robots on one plate pay only once the seams are split into non-interfering zones that still respect welding sequence.
Who this is for: Written for structural fabricators and project engineers laying out welding automation for large flat assemblies.
Scope: This EVST guide covers why large flat weldments are welded by a travelling gantry rather than a floor-mounted arm, what inverting the robots buys, and what has to be settled before two robots work one plate. It does not quote distortion figures, cycle times or interference clearances; those come from simulation and from your welding procedure.
The part stays where it is
Large flat weldments are awkward to turn and rarely worth turning. A single lift consumes crane capacity and time, and it introduces handling risk on a part that is often already tacked and dimensionally sensitive.
In the reference footage the plate assembly lies on a floor platform and does not move at any point. A gantry spans it and the robots travel to the seams instead.
That is the trade the layout makes: instead of bringing the part to the equipment, the equipment is brought to the part. It suits large, flat parts with seams distributed across the area, and it suits them poorly if the work needs frequent repositioning of the part itself.
What inverting buys on a gantry mounted welding robot
Hanging the robots under the crossbeam removes their footprint from the floor. Nothing stands beside the part, so the working area can be laid out to the limit of the gantry travel rather than around a set of bases.
The floor a base would have occupied is precisely the floor you want for the workpiece, which is why the saving is larger than the base area suggests.
Services follow the same logic. Cables, hoses and wire feed run overhead with the carriage instead of crossing the floor, keeping traffic lanes and crane access clear in a bay where both matter.
The cost is that maintenance access is overhead, and that the gantry structure itself has to be stiff enough that the arm’s accuracy is not spent on beam deflection.
Coverage is a travel figure, not a reach figure
The working range of this arrangement is set by the longitudinal and transverse stroke of the gantry. Arm reach only determines what postures are available once the carriage has arrived.
That reverses the usual sizing order. Stroke is fixed first, against the largest part the station has to take plus the access it needs; the robot is chosen afterwards for the postures the seams demand.
Doing it the other way round – picking a robot and then asking how far it has to travel – almost always produces rework, because the robot choice does not constrain the part size and the stroke does.
The order EVST works in is the same each time: part size, then gantry stroke, then mounting, then the robot, and seam zoning once those are fixed.
| Part and seams | Usual arrangement | What to verify first |
|---|---|---|
| Large flat plate, seams spread over the area | Travelling gantry, robots inverted | Stroke against the largest part, plus beam stiffness |
| Long part, seams along one axis | Robot on a linear rail beside the part | Rail length and access on the far side |
| Part that can be rotated to bring seams down | Fixed robot with a positioner | Positioner capacity and repositioning accuracy |
| High seam density on one large plate | Two robots on one gantry, zoned | Interference envelopes and welding sequence |
Two robots on one plate: zone first
The footage shows two robots welding the same plate at the same time. The benefit is immediate, but it is conditional on the seams having been divided into zones beforehand.
Three things have to hold in that division. The arms must not be able to reach the same space at the same time; their travel must not block each other on the shared structure; and the sequence must still make sense thermally.
Get the split wrong and the saving returns as waiting, as replanned paths, or as extra straightening work after welding. The first two are visible immediately; the third is not discovered until the part is measured.
According to ISO 13920:1996, general tolerances for welded constructions are defined as tolerance classes, which is the frame in which a distortion result is judged acceptable or not.
Heat input distribution, and what belongs to simulation
Welding two zones simultaneously changes how heat enters the assembly compared with welding them in sequence. That is a predictable consequence; the resulting distortion is not predictable from footage.
According to EN 1090-2:2018, execution of steel structures is specified by execution class, with the requirements for welding and for dimensional tolerances following from it, so the acceptable outcome is defined before the sequence is chosen rather than after.
Interference checking is a simulation task. The combined envelope of two arms on a moving gantry has to be verified point by point, and a video of two robots working without collision is not that verification.
Every EVST application review opens from the same project inputs – plate thickness, plate size and seam layout – so selection rests on evidence and acceptance is agreed against your own parts rather than a datasheet.
Safety around a travelling gantry
According to ISO 10218-2:2025, the safeguarded space follows from the movement of the whole integrated system. For this arrangement that includes the gantry travel and the carriage, not only the arms hanging from it.
According to ISO 12100:2010, overhead work, the travel path, arc radiation and access for loading and unloading the platform belong to the same risk assessment as the robot motion.
Because the part is loaded by crane onto the platform inside the same footprint the gantry travels over, load and unload interlocking is usually the part of the design that takes the longest to settle.
Frequently asked questions
Why not use floor-mounted robots with longer arms?
Because reach scales poorly against plate size, and every base occupies floor beside the part. A gantry converts the coverage problem into a travel problem, which scales with the structure rather than with the arm.
What does inverting the robots actually gain?
Floor area beside the workpiece and overhead service routing. The area a base would have taken is the area you want for the part, and keeping cables off the floor preserves traffic and crane access. The trade is overhead maintenance access and a stiffness requirement on the beam.
When are two robots on one plate worth it?
When the seams can be divided into zones that do not interfere mechanically, do not block each other’s travel, and still allow a sensible welding sequence for distortion. Without that division the second robot adds waiting rather than throughput.
What should we send to get a useful answer?
Plate thickness and size, the seam layout, the execution class or tolerance class you work to, and how the part is loaded onto the platform.
Project inputs for an application review
To have this checked against your own part rather than a generic layout, send:
- plate thickness and maximum plate size
- seam layout and total weld length
- execution class or tolerance class you work to
- how parts are loaded onto and removed from the platform
Send the plate thickness and the seam layout and the gantry stroke and zoning can be worked through against them. Related reading: robot welding workstation, vision guided welding system.