Nobody buys a duct line because of ventilation. But I have watched shops lose good welders to coughs that would not quit, and I have watched a plasma table shut down an entire bay because the fume extraction was undersized and the operator refused to run it. Environmental control in sheet metal fabrication is one of those things that pays for itself quietly: fewer sick days, cleaner paint jobs, fewer visits from the inspector, and machines that last longer because metal dust is not grinding through every bearing.
This article is the practical version: where the hazards actually come from in a duct fabrication shop, what capture equipment works, what the noise sources really are, and what a sensible budget looks like. I am not a safety engineer; treat this as shop-floor experience plus common standards, and get your local compliance person to check the details.
A plasma table cutting galvanized or stainless produces a fine metal-oxide fume, and cutting aluminum adds oxide particles you do not want in anyone's lungs. The volume of fume scales with cutting time, so a table running 6 hours a day needs real capture, not a door propped open. On our CNC plasma cutting machines we recommend either a downdraft bed or water table, and the choice between them drives everything downstream.
A water table kills fume at the source, better than 90 percent capture on straight cuts, and quiets the process noticeably. The cost is water maintenance, slag cleanup, and rust on the slats. A dry downdraft bed pulls air down through grates at roughly 1.5 to 2.5 cubic meters per minute per square meter of table. It handles any material, but the filter bank needs changing and the fan eats electricity all day.
Spot welding galvanized duct spits zinc oxide fume at the electrodes, and it adds up fast on a spot welder running hundreds of welds per shift. Metal fume fever is real and welders famously shake it off by Monday morning, which is exactly why it goes unreported. Capture at the gun or a bench-side hood with 0.5 meters per second capture velocity at the weld point handles most of it.
MIG work on stainless, which some shops do on special orders, is a different tier: hexavalent chromium has hard exposure limits everywhere. High-vacuum extraction or a filtered booth is the realistic answer, not a ceiling fan.
Grinding dross and deburring edges makes coarse dust that mostly falls nearby but still ends up in eyes and in machine slideways. Lockforming and beading themselves are not big dust sources, but they are big noise sources, which we get to below. The spot welder and the grinding bench deserve one extraction point each, and that covers the welding-dust side of a small shop.
Shops ask whether they need source capture, ambient filtration, or both. Here is the honest comparison:
| Method | Capture efficiency | Running cost | Best fit |
|---|---|---|---|
| Water table under plasma | Very high for fume | Low power, water upkeep | Plasma tables up to 2 m wide |
| Downdraft bed plus cartridge filters | High, ~95% when sized right | Fan power plus filter swaps | Heavy-use plasma, all materials |
| Articulated extraction arms | High if positioned, poor if lazy | Low | Welding benches, spot welders |
| Ambient air cleaners (HEPA/ESP) | Low per pass, works as backup | Filter cost, moderate power | Polishing overall air after source capture |
| General dilution ventilation | Low | Heating/cooling penalty | Baseline only, never the main plan |
The pattern that works: capture at the source for plasma and welding, then one or two ambient units to mop up. Shops that rely on ambient filtration alone always end up with a visible haze by mid-afternoon, and haze means exposure.
Hearing damage creeps up without anyone noticing, which is why regulators treat 85 dBA as the action level in most jurisdictions. A duct shop has several machines sitting right around or above that line:
Pittsburgh lockformers and beaders run 88 to 95 dBA at the operator position, worst during thin-gauge high-speed runs. Plasma cutting adds 85 to 95 dBA plus compressor noise. Grinders are the worst offenders at 95 to 105 dBA. Even the decloiler brake releasing is a repeated 80-plus dBA clack that nobody counts but everybody hears eight hours a day.
Engineering first, always. Rubber-lined sheet guides on lockformers drop 3 to 5 dBA and cost almost nothing. Keeping bead rolls sharp matters more than people think: a worn bead roll hammering instead of rolling adds several decibels and vibrates the whole frame. Mounting a beading machine on isolation pads helps less than vendors claim, maybe 1 to 2 dBA, but it is free floor space-wise.
Enclosing the plasma compressor in its own room or a lined box usually takes the biggest single chunk out of bay noise. Grinding is hardest: capture hoods help fume, but the noise needs either a grinding booth with absorption or, realistically, limiting grinding time and providing proper muffs.
Then administrative measures: rotate lockformer operators every two hours, enforce hearing protection zones with actual signage, and include a hearing test in annual checkups. None of this is glamorous. All of it is cheaper than a compensation claim.
For a small shop, one plasma table, two spot welders, a lockformer, a beader and a grinding bench, here is a realistic spread from the projects I have been around. A downdraft or water-table upgrade for the plasma table runs from a few thousand dollars for a water table retrofit to ten times that for a full downdraft bed with cartridge filtration. Two articulated arms on the welding benches are in the low thousands each. A pair of ambient air cleaners adds a mid-four-figure sum. The compressor enclosure is mostly plywood and mineral wool, so labor plus a few hundred in materials.
Total for a credible setup lands around the price of one decent single machine. Compare that to one experienced welder's annual cost, and the decision gets easy. Compare it to a workplace respiratory disease cluster and it is not a decision at all.
Two smaller sources round out the list because they generate complaints even though nobody files them under safety. Compressed air blow-off for cleaning parts, if run above 2 bar at the nozzle, turns settled dust back into airborne dust and adds a shrill noise component. Sweeping floors with compressed air is the worst habit in the industry; a vacuum with a HEPA filter does the same job without re-suspending everything you just captured. If operators insist on air, fit venturi-style air guns with chip guards and limit line pressure to the gun.
Oil mist appears where machines use way lubricant or where a slitter runs flooded coolant on thicker stock. It coats walls and ceiling near the machine within months, and it degrades belt grip on conveyor transfers. Enclosed machines with mist collectors handle it; open machines need a hood above the wet zone. The tell-tale is a sticky film on the inspection lamp above the machine. Clean it once and watch how fast it returns; that rate tells you the mist collector is undersized or off.
Put the noisy and dusty processes on the far wall from assembly and QC, sharing one extraction run. A plasma table and grinding bench back to back can share a filter unit if the fan is sized for combined flow with a damper per branch. Keep the shearing machine and folding stations in the clean zone; they generate neither much dust nor much noise and do not need extraction, so putting them near the office side keeps the extraction ducting short and cheap.
Leave roof or wall space above the plasma table for the extraction riser. Retrofitting a duct penetration through a completed roof is exactly as fun as it sounds.
Walk the shop at 10 a.m. on a production day with the lights off. Any haze visible in the light beams is airborne dust your capture is missing; note where it starts. Put a sound level phone app, imperfect but indicative, at each operator position for five minutes and note anything over 85 dBA. Ask each operator one question: what task makes you cough or ring your ears afterward. Three answers, three priorities. That list is your first investment order.
Yes, and it is actually one of the better options, since the water captures fume at the cut before zinc oxide becomes airborne. Change the water regularly because it accumulates fine sludge, and dispose of it per local rules. Keep the slats clean so the workpiece sits flat, or edge quality suffers.
Common design guidance is 1.5 to 2.5 cubic meters per minute per square meter of grate area, with the higher end for stainless or heavy cutting. Undersizing shows up as fume drifting up at the plate edges, so if you see haze escaping the perimeter, add a perimeter slot exhaust before you blame the filters.
Aim for under 85 dBA at the operator position. Realistically a busy lockformer sits at 88 to 92 dBA, so combine engineering fixes, sharp rolls, rubber guides, isolation mounts, with job rotation and hearing protection. Watch the impulsive noise from blanking and brake releases too, since those peaks matter to hearing even when the average looks acceptable.
For daily management, face velocity readings at hoods and a smoke-pencil check at capture points are enough to catch problems early. For regulatory compliance in most jurisdictions, and for any insurance or client audit, you will need a documented survey with calibrated instruments, usually every one to three years. Keep the maintenance logs on filter changes; inspectors ask for them first.