Collaborative Welding on an Open Bench
Direct answer: Collaborative welding is often deployed on an open bench, which is a reasonable arrangement for low volume work and also a design obligation. Arc light, fume and heat travel differently and cannot be covered by one measure. Start from how close people actually get during loading, wire changes, slag removal and inspection, then size screening, extraction and thermal controls, and discuss cycle time last.
Who this is for: Written for engineers deploying collaborative robots on welding work, where the bench sits in a shared area rather than inside a fenced cell.
Scope: This EVST guide covers why the three welding hazards propagate differently, why the starting point is human approach distance, and what a risk assessment needs as input. It does not perform a risk assessment, quote exposure limits, or evaluate any specific installation.
Collaborative welding on open benches
In the reference footage a collaborative arm welds directly on an open steel bench, with a welding power source, a wire feeder and gas cylinders beside it and open floor around.
Small batch fabrication frequently looks like this, because a compact arm on a bench is quick to redeploy and does not consume the floor area a fenced cell needs.
That arrangement is a design decision with consequences, not a shortcut. What it removes is the physical boundary that would otherwise contain what welding produces, which makes collaborative welding a selection question about controls rather than about the arm.
Collaborative operation is about the application
ISO/TS 15066:2016 provides guidance for collaborative robot applications, and ISO 10218-2:2025 assesses the application rather than the arm. Neither treats a robot as safe by category.
A collaborative arm welding on a bench is a welding application that happens to use a collaborative arm. The hazards that dominate are the process hazards, not the contact hazards the arm’s own functions address. According to ISO/TS 15066:2016, guidance is given for collaborative robot applications rather than for robots in isolation, which is the same distinction collaborative welding runs into.
That distinction matters when a supplier offers a collaborative arm as an answer to safety. It answers one class of hazard.
Three hazards, three propagation paths
Arc light radiates. It affects anyone with line of sight, at distances well beyond the reach of the arm, which is why screening is positional rather than perimeter based.
Fume rises and disperses. Capture works close to the source; general ventilation dilutes but does not capture, and the difference shows up in what a worker actually breathes.
Heat stays with the mass. Workpiece, fixture and spatter remain hot after the arc stops, so the control is procedural and temporal as much as physical.
One measure sized for one of these does little for the other two.
Start from approach distance
The useful first question is not which guard to buy but how close a person gets, and when. Loading and unloading, wire changes, slag removal, nozzle cleaning and inspection each bring someone to a different distance for a different duration.
Mapping those tasks produces the distances that screening and extraction have to be designed against. It also exposes tasks nobody had counted, which is often where the real exposure sits.
According to ISO 11611:2015, protective clothing for welding and allied processes is specified for the process conditions, which makes it one layer among several rather than a substitute for controls at source. According to ISO 12100:2010, protective measures follow after the limits and the tasks have been determined, and acceptance of a collaborative welding station should be written against those tasks.
What the footage proves and what it does not
It proves that a collaborative arm strikes and runs a welding arc on an open bench, in view of the surrounding area, with standard welding equipment beside it.
It does not prove whether that installation has been risk assessed, what controls exist outside the frame, or whether it meets any standard. A single camera angle cannot establish the absence of anything.
This article is about the questions this class of installation has to answer. It is not an assessment of the cell in the footage, and no acceptance conclusion should be drawn from it.
A short decision table
Nobody has line of sight during welding and nobody approaches until the part has cooled: screening and thermal controls are simpler, extraction still has to capture at source.
People work nearby throughout: screening becomes positional, extraction has to follow the torch, and task timing has to be designed rather than assumed.
Frequent manual intervention between welds: the procedural controls dominate, and cycle time should be set after they are fixed.
| Approach pattern | Dominant control | Set before cycle time |
|---|---|---|
| Nobody in line of sight while welding | Capture at source for fume | Extraction position |
| People working nearby throughout | Positional screening | Screening and extraction together |
| Frequent manual intervention between welds | Procedural and thermal controls | Task timing and cool-down |
What to prepare before asking for a quotation
Workpiece and joint information: material, thickness, joint form and weld length. Station layout with people’s routes and load and unload positions marked.
Existing ventilation and extraction, and whether capture at source is possible where the arc sits.
Task list with approach distances and frequencies, cycle requirement and shift pattern.
Frequently asked questions
Does a collaborative robot remove the need for guarding on welding?
No. It addresses contact hazards. Arc light, fume and heat are process hazards and need their own controls whatever arm is used.
Is general workshop ventilation enough for welding fume?
Dilution is not capture. Capture works close to the source, which is why extraction is usually positioned at the arc rather than at the roof.
Where should a risk assessment start?
With the tasks people actually perform and how close they get, including maintenance and intervention tasks, not only normal production.
Can cycle time be quoted before the controls are decided?
It can, but it usually changes once screening, extraction and intervention timing are settled, so it is worth settling those first.
Project inputs for an application review
If you are deploying collaborative welding on an open bench, the task and approach map is what turns a quotation into a design. EVST starts from the task list rather than from a guarding catalogue.
- workpiece, joint form, thickness and weld length
- station layout with people’s routes marked
- existing ventilation and whether capture at source is possible
- task list with approach distances, frequencies and shift pattern
Send those and EVST will work through layout and controls against your actual station. Related reading: welding robot workstation, smart robotic factory.