Key takeaways
- Cobots earn their keep on duct runs with repeated rivet patterns, not one-off custom geometry.
- Force sensing and a stable buck side matter as much as cycle time for airtight joints.
- Present parts at a fixed datum so teach points survive low-volume product changes.
- Labor pressure in sheet metal trades makes even partial automation worth piloting.
When does a cobot belong on a duct riveting station?
A collaborative robot fits HVAC duct riveting when the same joint pattern repeats often enough that a fabricator's wrist and shoulder time becomes the bottleneck. Rectangular sections with standard flange layouts, collar-to-boot assemblies, and long straight runs are typical wins. Custom architectural transitions with odd angles still belong with a skilled fitter at the bench.
The opening case is not full-line automation. It is one station where an operator loads a section, triggers a taught rivet sequence along a known edge, and moves to deburr or seal while the arm handles the percussive work. If your shop still loses hours hunting missing rivets and swapping mandrels on every size change, fix material flow first. A cobot on a chaotic bench only automates the noise.
Industry reporting on sheet metal trades points to persistent hiring pressure, with a large share of contractors citing shortages and more workers nearing retirement than entering the trade. That backdrop pushes duct shops toward any repeatable task a cobot can hold steady without adding another full-time riveter.
Which duct joints are realistic to automate first?
Start with joints where hole spacing, flange overlap, and buck access are predictable. Longitudinal seams on straight duct, repeated collar patterns, and accessory tabs on standardized fittings are easier than tapered transitions that need constant shimming.
Many shops already move toward clinching or lock-forming on high volume lines to cut rivet consumables. Cobot riveting still matters where rivets remain the specified fastener on a job, on retrofit panels, or on mixed lines that cannot justify a dedicated clinch cell. The robot should mirror the joint your spec sheet already allows, not invent a new stack-up.
Match the automation level to batch size. A night run of fifty identical sections pays teach time back fast. Ten completely different one-offs in a day rarely does unless programs are stored and recalled with only offset tweaks.
How should you present parts to the arm?
Duct sections are large, floppy, and easy to twist out of tolerance. A cobot cell needs a presentation fixture with hard stops on the flange plane and at least two datums along the seam edge. Vacuum or mechanical clamps at the buck side beat an operator trying to hold a 10-foot section square while the gun fires.
Reach limits matter. Mount the arm so the rivet path stays inside the collaborative workspace without the wrist folding past comfortable angles. Overhead clearance for tall sections may push you toward a sliding fixture that indexes the work instead of stretching the arm.
Keep the human zone for fit-up and inspection separate from the rivet path. Light curtains or area scanners are common in fab shops where someone steps in to check square. Document those pauses in the risk assessment before you run unattended cycles.
What role does force control play in rivet quality?
Percussive riveting sends shock through the gun, the workpiece, and whatever backs the shop side. Research on human-robot collaborative riveting shows that compliant control and force sensing improve consistency compared with stiff position-only moves, especially when the buck bar must react to each strike.
On duct gauge metal, overdriving a rivet can oval the hole or tear thin galvanized skin. Underdriving leaves a loose shop head that fails leak checks. Program with force limits and optional search moves at the buck station so the arm finds contact before the gun cycles.
Vibration damping on the buck side is not cosmetic. Studies comparing bucking bar designs report large reductions in peak squeeze force transmitted to the arm when softer interfaces are used. That protects both the cobot joint seals and the operator who still loads parts during the shift.
How do tooling and changeovers work in high-mix shops?
HVAC fabrication rarely runs one diameter all week. Quick-change rivet sets, color-coded mandrels, and a posted chart at the cell cut minutes off every size swap. Store teach programs by flange type and sheet thickness, not by customer name, so recall stays searchable.
Batch similar work before you reteach approach angles. If collar riveting and straight seam riveting need different end effectors, budget two tool plates on a manual changer rather than forcing one gripper to do everything poorly.
Offline touch-up should stay rare. When a new fitting arrives, have the programmer capture one golden section, verify with a pull or leak spot check, then lock the program version in your library.
How do you verify quality without slowing the line?
Rivet quality on duct is judged by shop head form, flush field side, and whether the joint holds vacuum or smoke test when the spec requires it. Build verification into the cell with go-no-go gauges at the buck station and a periodic audit lane where a lead pulls one joint per batch.
Vision can count rivets along a seam for missing strikes, which catches a skipped index better than eyeballing a 12-foot run. Keep vision on presence and position, not cosmetic paint, unless you have stable lighting on galvanized sheet.
Log strike count and program ID per section if you trace jobs for commercial work. When a callback happens, you want to know which teach file ran, not guess from memory.
What does a sensible pilot look like?
Pick one flange family, one rivet size, and one customer with steady repeats. Run two weeks manual baseline on cycle time and rework, then two weeks with the cobot handling only the straight rivet path while humans keep fit-up.
Track injuries and fatigue complaints on the rivet station, not just parts per hour. Even a modest cycle gain matters if it stops a second shift hire you cannot fill.
Service Robot Co. can scope a vendor-neutral cobot rental or lease path, map safety around your existing fab layout, and connect you to nationwide field service when a tool changer misaligns mid-week. The goal is a measured cell, not a brochure robot.
What should you ask before you sign?
Ask how the integrator handles percussive loads on the arm rating and whether the buck fixture is included in the lease. Ask for sample cycle data on your gauge and flange, not on demo sheet.
Ask who owns program storage when you add a second brake or shear upstream. Ask how fast you can get a technician if the force sensor drifts after a heavy production week.
Ask about month to month cobot rental terms if you want to prove the cell through a busy season before capitalizing. Financing flexibility matters when duct schedules spike with construction starts and slow with weather.
