Welding Fume Extraction Nozzles: Fix Capture Velocity, Distance, and Cross Draft

A cell runs a 250 A MIG seam, the extraction arm gauge reads 1,400 m3/h, and visible fume still rolls past the hood lip into the operator’s breathing zone an hour after the arc strikes. Published industrial hygiene surveys keep finding the same pattern: the exhaust hardware is often sized correctly, and the capture zone is simply in the wrong place. A weld plume is a weak buoyant jet rising at roughly 0.25 to 1 m/s for most manual processes, so even mild room movement can push it outside the hood before any captured air reaches the duct. If your local exhaust ventilation (LEV) cannot beat that plume at the point of generation, the fan curve is irrelevant.

This article is a capture-efficiency walkthrough for welding fume extraction, framed entirely with published ranges from the ACGIH Industrial Ventilation Handbook and standard airflow arithmetic. No health claims, no vendor data, just the numbers that decide whether fume enters a hood or the room. When a welding fume extraction nozzle ring replaces brute-force airflow, the same cell can drop from thousands of cubic meters per hour to a few hundred, and that shift is worth understanding in detail.

The Snapshot

• Required capture velocity spans 0.5 to 1.0 m/s for gently released fume in still air, and 2.5 to 10 m/s where releases are energetic or cross drafts are strong, per the ACGIH Industrial Ventilation Handbook range tables.
• A 300 mm hood face at 0.75 m/s moves about 190 m3/h. Demand that same 0.75 m/s at 1 m from a plain opening and the handbook estimate Q = V(10x2 + A) jumps to roughly 27,000 m3/h.
• A steady 0.5 m/s cross draft tilts a typical arc plume 30 to 60 degrees from vertical, which routinely doubles the airflow needed to hold the capture zone on a side hood.
• An annular jet ring with a 15 to 30 m/s welding fume extraction nozzle stream entrains 3 to 6 times its own air volume, rebuilding a 400 to 600 mm capture aperture at close range with 40 to 80 m3/h of supply air.

Why Capture Fails Before Fan Sizing Does

Fume generation is a small thermal engine. The arc heats surrounding air, the plume rises, and ambient air is dragged along with it. Published measurements place buoyant plume velocities near the source at 0.25 to 1 m/s for GMAW and SMAW, with submerged arc and high-current processes occasionally exceeding 2 m/s. Particle counts concentrate below 1 micrometer aerodynamic diameter, so the fume follows room air almost perfectly and has no inertia to fight drafts.

A side hood captures only where its induced velocity exceeds the plume velocity plus whatever disturbance the room adds. At 500 mm from a 200 mm round extraction arm inlet, centerline velocity is already a fraction of the face value, and the geometry decay discussed below does it no favors. Three failure modes account for most weak capture: too much distance, uncontrolled cross draft, and a hood shape that wastes the airflow it does receive. A fourth, fan undersizing, is rarer on installed arms because arms ship with matched blowers. The remedy starts with arithmetic.

The typical shop device for manual welding is a 200 mm extraction arm with an integral blower, or a fixed hood on a backdraft or downdraft table. Published arm specifications commonly quote 850 to 1,700 m3/h free airflow at 3 to 7 kW of fan input, with capture effectiveness concentrated inside a 300 to 500 mm working radius from the inlet. Within that radius, a flanged arm inlet behaves like a close-coupled hood at 0.5 to 1.0 m/s capture velocity, which is exactly the still-room range from the table later in this guide. Outside that radius, treat the arm as decorative and size the real capture device separately.

For backdraft and downdraft benches, the governing number is face velocity through the work-plane openings, typically 0.5 to 1.0 m/s at the slots, with room exhaust of 15 to 25 air changes per hour for a 3 to 4 m ceiling bay. A 4 x 2 m downdraft table with 20 percent open perimeter slots needs about 4 x 2 x 0.2 x 0.75 x 3,600, roughly 850 m3/h just for the slots, before the weld plume adds anything. Arm, hood, and table are three different devices with three different arithmetic. Choose the one whose capture geometry actually reaches the arc.

Capture Velocity Ranges From Published Guidance

The ACGIH manual publishes control velocity ranges by draft condition rather than fixed numbers, because the room decides the requirement. The table below restates those widely cited ranges and converts them into exhaust examples for a 300 mm circular collection face (area 0.0707 m2, close-coupled, using Q = V x A).

Draft condition at the weld	Required capture velocity	Example exhaust, 300 mm face
Gently released fume, still room	0.5 to 1.0 m/s	130 to 250 m3/h
Normal shop activity, light air movement	1.0 to 2.5 m/s	250 to 640 m3/h
Release with turbulence, doors or carts nearby	2.5 to 5.0 m/s	640 to 1,270 m3/h
High cross draft, energetic release, gantry travel	5.0 to 10 m/s	1,270 to 2,550 m3/h

Read the middle column as the velocity you must induce at the plume, not at the fan. The right column assumes the hood sits close-coupled to the source, one hood-width or less from the weld. The moment distance grows, the same velocity costs multiples of that air. A cell that needs 250 m3/h at 150 mm may need several thousand at 1 m, which is why arms mounted “out of the way” quietly stop working.

The Distance Penalty On A Plain Opening

Air pulled into an unflanged opening comes from all directions, and velocity along the centerline decays with the square of distance. The ACGIH estimate for a plain opening is Q = V(10x2 + A), with Q in m3/s, x in meters, and A the hood area in m2. Work the required example: a 300 mm hood (A = 0.0707 m2) holding 1.0 m/s capture velocity at x = 1 m needs Q = 1.0 x (10 x 1 + 0.0707) = 10.07 m3/s, about 36,000 m3/h. That is a roof-scale exhaust for one weld, and it is still fragile.

Halve the distance and the arithmetic collapses: at x = 0.5 m the same expression gives roughly 2.6 m3/s, about 9,300 m3/h. Add a flange or a deep enclosing hood and the 10x2 term largely drops out, leaving approximately Q = V x A. This is the single most important rule in fume extraction: capture within one hood-width of the source, or stop pretending the hood is capturing. When geometry forbids close coupling, the engineering answer is not a bigger fan. It is moving the capture surface toward the plume, or dragging the plume toward the surface with a jet.

Cross Draft At 0.5 M/s Rewrites The Design

Cross draft is the quiet killer of every LEV design. Forklift wakes reach 1 to 2 m/s, open bay doors pass steady 0.3 to 0.8 m/s streams, overhead destratification fans and ceiling HVAC diffusers commonly deliver 0.25 to 0.5 m/s right through the breathing zone. Published guidance treats room velocities near 0.5 m/s as the practical limit for conventional hoods; above it, the fume path bends away faster than the hood can chase it. A plume tilted 45 degrees lands outside a side hood that captured cleanly under still air.

Diagnose before you buy: a smoke pencil or a glycol fog wand at the arc shows the failure in seconds. Mitigation is ordered by leverage. Block the draft with a temporary screen or draft wall within 1 m of the weld. Lower the supply so it does not sweep the work zone. Move the hood to windward of the plume. Or abandon passive capture and inject momentum of your own, which is where nozzle-based methods enter. A fixed welding fume extraction nozzle ring is largely immune to mild drafts because its jet velocity at the capture aperture is 10 to 60 times the plume velocity, while a hood relying on induced suction at distance is not.

Hood Geometry And The 40 Degree Rule

Hood shape sets how much of your airflow survives entry. For conical enclosing hoods, practice limits the half angle to 40 degrees or less, keeping total included angles near or below 90 degrees. Shallower tapers hold entry loss coefficients around 0.10 to 0.17; steeper cones push toward 0.25 and above, shedding static pressure as swirl and separation while fume re-entrains off the hood wall. ACGIH entry loss tables carry the same message for offset and plain hoods: turn the air gently or pay for it in fan pressure.

Slot geometry multiplies the gain. A 100 x 300 mm rectangular collector mouth behaves like a much wider hood while keeping air volume low, because velocity concentrates in the slot. A flange of 0.5 to 1 hood-diameter around a plain opening cuts the required flow by suppressing radial induction from the back side. Check three numbers on any hood you already own: included angle, flange presence, and standoff distance. If the hood is steep, unflanged, and 700 mm away, it is a decorative object, and the next two sections show the repair options.

Jet Induction: The Nozzle Ring That Pulls The Plume

A jet-induced capture ring puts a welding fume extraction nozzle array where the plume is born. Small nozzles, typically 1 to 3 mm bores arranged on a 100 to 200 mm circle around a torch or tool holder, fire an annular sheet of air at 15 to 30 m/s. The jet shears against still room air and drags it forward; that entrained flow forms a coherent low-pressure column that swallows the plume instead of waiting for suction to reach it. Representative momentum-jet literature puts entrainment at 3 to 6 times the nozzle supply volume within 10 to 20 diameters of the exit, with centerline velocity decaying roughly as 1/x along the axis.

The practical numbers are attractive. A six-nozzle ring at 0.3 m3/s total, roughly 65 m3/h of compressed air at 2 to 4 bar, can hold a 400 to 600 mm capture aperture at 200 to 400 mm standoff. A plain hood doing the same duty from that distance is a four-digit airflow, per the section above. Downstream, the ring exhausts into a close-coupled collector or the cell’s duct and fan, which can now be resized much smaller. Selection of the nozzles themselves follows normal compressed-air jet rules covered in our air nozzle selection guide, and the hardware category is air nozzles. Watch two limits: jet angle must intersect the plume path, and supply air adds heat and moisture noise to the cell.

Air Curtains As Virtual Hoods

An air curtain is a continuous flat sheet of air, usually from a flat fan nozzle header or an air knife, stretched across a cell opening. It does not capture anything directly. It walls the plume in. A slot discharging 10 to 20 m/s with 20 to 50 mm of opening depth produces a sheet that stays coherent across 1 to 2 m of span and kills the 0.5 m/s cross drafts that break conventional hoods.

Typical welding layout: a curtain header above the cell opening tilts 10 to 15 degrees toward the extract side, closing the breathing-zone face of the work envelope while the extraction arm or slot hood handles suction behind the sheet. This converts an open bay into a quasi-enclosure for a few hundred euros of nozzle and duct. The sheet itself should target 1 to 3 m/s coverage velocity at the opening plane. For the mechanics of flat sheet versus point jets in blow-off duty, see the comparison in flat fan air nozzle blow-off; the same velocity-decay logic governs curtain design.

Displacement Supply And Room-Level Strategy

Displacement ventilation takes the opposite route: stop fighting the plume and stabilize the room instead. Low-level supply air, typically 15 to 20 degrees cooler than the space, is admitted at 0.2 to 0.5 m/s through floor grids or bench-level diffusers. A cool lake forms across the working plane, plumes rise undisturbed through it, and high-level extract removes them. This is standard practice in robotic welding bays and it pairs well with low-speed conditioning jets that do the same gentle, sub-0.5 m/s delivery work that misting systems rely on, adjusted for shop heat loads.

Three numbers decide whether displacement works: supply temperature differential of 2 to 5 K, floor velocity under 0.25 m/s at the work plane, and extract inlets positioned above the thermal plume height, usually 2.5 to 4 m. It fails when forklift traffic or open doors at work-plane height overwhelm the buoyant layer, typically above 1 m/s. Displacement fixes room-level transport. It does not replace point capture at the arc, and no serious design relies on it alone for a manual station.

Enclosing Laser And Robotic Welding Cells

Laser welding fume behaves differently from arc fume. Keyhole processes eject vapor at meters per second with high ultrafine content, and published laser studies treat the plume as a directed jet rather than a lazy buoyant column. That velocity is leverage: a small enclosure and a modest exhaust will take it. Robotic cells go further because the torch path is repeatable, so the capture geometry can be hard-fixed.

A practical robot-cell recipe from published practice: seal the cell to a leakage target of 25 to 50 Pa negative pressure, size the exhaust for 10 to 20 air changes per hour or for 0.5 to 1.0 m/s face velocity across every opening, whichever is larger. Add a purge interlock so extraction runs 30 to 60 seconds before and after the arc. Fit a nozzle ring or slot collar at the lens housing, 50 to 150 mm from the weld, because laser plume capture rewards very close coupling. For fiber lasers above roughly 4 kW, plume velocities at the interaction zone routinely exceed 10 m/s within millimeters, then decay fast, so the collar sits inside the closest few jet diameters. Our dust suppression nozzle guide covers the collar-versus-enclosure trade in particulate duty, and the same logic transfers.

Noise And Energy Trade-Offs Of Brute Airflow

Oversizing feels safe and is expensive. Fan laws punish it: with a fixed duct system, pressure demand scales with Q squared and shaft power with Q squared-times-pressure, so doubling airflow multiplies absorbed power by roughly eight. Sound follows too, adding about 9 dB per factor-of-eight power step. A cell that “needs” 3,000 m3/h because a hood sits too far away is a 10-times-energy cell, and its blower plus arm noise pushes the operator toward hearing-protection territory at 1 m.

The jet-induction alternative trades fan kilowatts for compressed air and local noise. A 65 m3/h nozzle ring at 3 bar draws on the order of 1 to 2 kW of compressed-air power, while the downsized hood fan drops from several kilowatts to well under one. But an undamped ring is a whistle: open jets above 25 to 30 m/s can produce 75 to 85 dB(A) at 1 m, so engineered low-nozzle-noise designs and enclosures matter, following the same amplified-flow principles we cover in air nozzle for air compressor. Target 80 dB(A) or less at the nozzle plane. If the math lands within a factor of two either way, choose the quieter option; noise retrofit costs more than airflow retrofit.

Commissioning And The Smoke-Pencil Test

Capture is a physical condition, not a specification, so commission it as one. Walk the finished cell with a smoke pencil or fog generator at the arc for ten minutes. Fume should turn toward the capture aperture within 0.3 m and never escape past the hood edge or curtain sheet at any weld position. Measure face velocity at the hood plane with a thermal anemometer and accept plus or minus 20 percent of design.

Then stress it. Open the bay door, run a forklift pass, set destrat fans to high. If the plume escapes under any of those, you have sized a hood for a room that does not exist. Fixtures to recheck after every process change: standoff distance, hood angle, flange seating, nozzle ring alignment, and filter loading, because a filter at twice its clean pressure drop quietly removes a third of your capture airflow. Document the smoke traverse; auditors treat it as real evidence, and so should you.

A short field kit makes the whole loop repeatable in under an hour. You need a thermal anemometer with a 0.15 to 20 m/s range, a telescoping probe for hood faces and ring apertures, a smoke pencil or glycol fog wand, a tape measure, and a manometer or magnehelic gauge for the duct static tap. Log four numbers per station every quarter, in one line of a spreadsheet: hood diameter, standoff in millimeters, face velocity in meters per second, duct static in pascals. Capture failures announce themselves as drift in those four columns long before anyone complains about visible fume, and the drift is usually distance and filters, not motors.

That log feeds a ten-minute sizing review you can run per station. Stand at the weld and time the plume tilt with a smoke pencil; if it leans more than 15 degrees off vertical, your draft condition is the middle row of the table, not the top. Measure the maximum standoff your fixture allows; if it exceeds one hood-width, plan for a jet ring or a flanged close-coupled collector rather than a plain opening. Compute Q = V(10x2 + A) at the required velocity, then the flanged close-coupled alternative, and compare both with what the arm delivers at the dirty-filter end of its life, not the clean end.

Then pick the intervention with the steepest return. Distance reduction is nearly free and multiplies capture by 5 to 20 times. Flanges cost 50 to 150 euros and recover the radial-induction losses. Draft screens cost a sheet of perforated steel and remove the cross-draft penalty at the bench. Nozzle rings cost compressed air, 40 to 80 m3/h at 2 to 4 bar per ring, and buy immunity to the rest. Only after those steps does fan capacity enter the conversation, and usually the answer is that the fan was never the problem.

When Arithmetic Points To A Nozzle Fix

Step back through the decision chain. Compute required capture velocity from the draft condition. Compute the plain-hood airflow at your actual standoff. If the answer is more than roughly three times what the arm can deliver, stop shopping for blowers. Close the distance with a flanged collector, wall the bay with an air curtain, or install a jet ring so the plume is dragged, not awaited. Each of those moves changes the physics instead of paying for the same physics at higher volume.

If you want the jet-ring route scoped, the variables are nozzle count, bore, supply pressure, and ring geometry relative to the torch, and a 15-minute conversation with your arc process details beats an afternoon of catalog guessing. Send the joint geometry and draft conditions through our contact page and we will walk the entrainment math with you before anyone quotes hardware. A welding fume extraction nozzle is a cheap component and an expensive lesson, and the lesson is always this: the plume loses every distance fight, so design as if it will.