Industrial Ventilation, Fans & Dust Collection: The Complete Engineering Guide — Capture Velocity That Sizes a 1,240 m³/h Welding Arm, Darcy–Weisbach Worked Through a 10,000 m³/h Shop System, the Fan Affinity Laws and the 58 % Energy a VFD Finds in a Fixed-Speed Extraction Fan, Cyclone Cut Diameters Near 8 µm Against Cartridge Collectors at 333 m² of Media, the Combustible-Dust Arithmetic Where Kst 450 Titanium and Kst 140 Wood Size Different Vents, Welding Fume at IARC Group 1 with Every Exposure Limit Written Down, and a 2026 Rupee Ledger from ₹23,000 Extraction Arms to a ₹2.2 Lakh-a-Year Energy Swing

Industrial ventilation and dust collection in numbers: capture and transport velocities worked to duct sizes, Darcy-Weisbach and fan affinity laws with a VFD energy case, cyclone, cartridge and baghouse sizing, combustible dust Kst and NFPA 660, welding fume and silica exposure limits, and a 2026 rupee ledger for Indian shops.

·

Industrial Ventilation, Fans & Dust Collection: The Complete Engineering Guide

This series has spent a year taking machines apart — boilers and bearings, pumps and PLCs, conveyors and cleanrooms — and each guide ended in the same place: a ledger, because on a factory floor the only argument that survives is arithmetic. This guide is about the one machine every factory owns without having ordered it: the air. It is the working fluid of every process that cuts, grinds, welds, paints, prints, or powder-handles anything — and in most Indian fabrication shops it is the last system to be engineered and the first to be neglected. The dust collector is duct-taped to the wall, the ducting was sized by the guy who sold the fan, and the fan itself runs at full speed year after year because nobody has ever computed what that costs. All three of those choices are expensive, and all three are fixable with the mathematics below.

The stakes are not decorative. Welding fume — the aerosol this guide keeps returning to — was reclassified by the International Agency for Research on Cancer in 2017 from "possibly carcinogenic" to carcinogenic to humans (Group 1), on sufficient evidence for lung cancer, and an estimated 11 million people hold the job title of welder while some 110 million more incur welding-related exposure [6]. Respirable crystalline silica kills tens of thousands in India annually through silicosis and its sequelae, with prevalence in stone workers measured at 12–30 % even where the national exposure limit is nominally met [21]. Combustible dust explosions continue to level foundries, wood shops and — since additive manufacturing put fine titanium and aluminium powders into ordinary workshops — a new generation of small facilities that never knew the hazard existed [17][18]. And through all of it runs the arithmetic in this guide: a welding arc that emits 312 mg/min of fume needs 18,700 m³/h of dilution air to reach a 1 mg/m³ breathing zone, or a 1,240 m³/h extraction arm pulling on it 300 mm from the source. The physics does not care which one you install; the rupee ledger very much does.

What follows, section by section:

1. The Contaminant and the Law

1.1 Five things air can carry

Industrial ventilation fails most often at the vocabulary stage, because "dust collection" is treated as one problem when it is five:

Contaminant · Definition · Typical source in a fab shop · Size range · Collection equipment

Dust · Solid particles generated by mechanical action · Grinding, sanding, cutting, blasting, powder handling · 1–500 µm · Cyclone + fabric/cartridge; HEPA for fines

Fume · Solid particles from condensation of vapour (metal boiled, then frozen) · Welding, laser & plasma cutting, foundry pouring · 0.005–20 µm, mass mostly < 1 µm · Cartridge/HEPA; never a cyclone alone

Mist · Liquid droplets · Machining coolant, spray painting · 1–50 µm · Mist eliminators, electrostatic, media

Gas / vapour · Molecules, not particles · Solvents, CO, ozone, NOx from arcs · molecular · Ventilation, activated carbon, source substitution

Fibres · High-aspect-ratio solid particles · Grinding of composites, insulation work · 1–100 µm long · HEPA-grade, special handling

The physics that decides everything downstream is particle size, because it sets three curves at once: how long a particle stays airborne (a 10 µm particle of density 2,000 kg/m³ settles at about 6 mm/s — roughly 10 minutes to fall three metres in still air; a 1 µm particle settles 100× slower and never really lands while air moves), how deep it penetrates a human lung (particles below about 4 µm aerodynamic diameter reach the alveolar region, which is why "respirable" fractions exist as separate exposure limits), and how hard it is to filter (filtration efficiency has a minimum near 0.1–0.3 µm — the MPPS region — so the submicron fume that looks least impressive is the hardest to catch).

Welding fume is the perfect illustration of why "it looks like smoke, it will be fine" is wrong. Fresh fume particles nucleate at 10–23 nm, coagulate within seconds into the 0.1–0.4 µm range, and a shielded-metal-arc plume's dominant mode — 80–90 % of all particles — sits at a geometric mean diameter of 0.22 µm, squarely in the window that sails through a coarse filter, through the nasal passages, and into the gas-exchange region of the lung [9][10][11]. Only 10–30 % of the fume mass is bigger than 1 µm [9]. Everything this guide says about filtration, capture velocity and hood placement follows from those two numbers.

1.2 The law (the part most shops never read)

Indian factories law has required exactly what this guide will compute since 1948. The Factories Act, §14 — "Dust and fume":

"In every factory in which, by reason of the manufacturing process carried on, there is given off any dust or fume or other impurity of such a nature and to such an extent as is likely to be injurious or offensive to the workers employed therein, or any dust in substantial quantities, effective measures shall be taken to prevent its inhalation and accumulation in any workroom, and if any exhaust appliance is necessary for this purpose, it shall be applied as near as possible to the point of origin of the dust, fume or other impurity, and such point shall be enclosed so far as possible." [20]

Read the bolded clause twice, because it is an engineering instruction disguised as a legal sentence. It says: extract at the source, and enclose it if you can. That is precisely the order of preference — enclosure beats local exhaust beats dilution — that industrial ventilation has rediscovered over seventy years of field measurement, and it is the cheapest sentence in this entire guide to obey. Section 13 of the same Act adds the ventilation-and-temperature duty; the state factory rules and DGFASLI circulars carry the enforcement details; and the same principle appears in every modern code — ACGIH's Manual of Recommended Practice, ASHRAE's industrial local-exhaust chapters, and the US OSHA rules that give the numbers a price [1][2][3].

1.3 The numbers that give the law teeth

Exposure limits come in three flavours, and knowing which is which prevents expensive mistakes. PELs (OSHA) are legally enforceable; TLVs (ACGIH) are health-based recommendations, typically stricter, and in most countries — and in most competent industrial-hygiene practice — the TLV is the number you design to; RELs (NIOSH) are research-based recommendations. The relevant set for a fabrication shop:

Contaminant · OSHA PEL · NIOSH REL · ACGIH TLV · Notes

Welding fume (iron/mild steel/aluminium total) · 5 mg/m³ · — · 5 mg/m³ (legacy) · Applies to total fume; specific metals get their own limits on top

Manganese (fume/compounds, as Mn) · 5 mg/m³ ceiling · 1 mg/m³ TWA; 3 STEL · 0.02 mg/m³ respirable; 0.1 mg/m³ inhalable · TLV cut from 0.2 in Dec 2012; CNS effects are the basis

Hexavalent chromium Cr(VI) · 5 µg/m³ TWA (also an Action Level of 2.5) · 0.2 µg/m³ · 0.0002 mg/m³ = 0.2 µg/m³ TWA; STEL 0.5 · Group 1 carcinogen; OELs are microgram-scale

Respirable crystalline silica · 0.05 mg/m³ TWA (2016 rule) · 0.05 mg/m³ · 0.025 mg/m³ · India DGMS limit for mines: 0.15 mg/m³ — 3–6× looser [21]

Wood dust (hard & soft) · 15 mg/m³ total; 5 respirable · — · 1 mg/m³ (many species; hardwood stricter) · IARC Group 1 carcinogen; explodeable (see §7)

Total dust / PNOR (nuisance) · 15 mg/m³ · — · 10 mg/m³ inhalable · The "generic" limits apply only where no specific limit exists

Two features of that table organise the whole design problem. First, the gap between the total-fume limits and the metal-specific TLVs is enormous: a shop that holds exactly 5 mg/m³ of stainless-steel welding fume is at roughly 25–75 µg/m³ of manganese — against a respirable TLV of 20 µg/m³ — and simultaneously tens of µg/m³ of Cr(VI) against a limit of 0.2 µg/m³ [5][6]. Complying with the visible limit confers no protection against the invisible one. Second, the microgram-scale Cr(VI) limit makes dilution ventilation arithmetically impossible for stainless work (§8), which is why process substitution and source capture are the only strategies in existence for it — exactly what §14 of the Factories Act ordered in 1948.

2. Capture: The Hood Is an Air Pump With Terrible Manners

2.1 Why dilute, when you can capture

There are exactly three ways to keep a contaminant out of a worker's breathing zone:

  1. Dilute — supply and exhaust so much room air that the contaminant concentration falls below the limit (general or dilution ventilation).
  2. Capture — pull the contaminant into a hood at the point of generation, before it joins the room air (local exhaust ventilation, LEV).
  3. Enclose — put a box around the process so the contaminant never enters the room at all (enclosing hoods; dust-tight cabinets).

The hierarchy is not a preference, it is a consequence of arithmetic. A rubber-tired industrial problem: capture velocity decays with the square of distance from the hood. For an ideal flanged circular openings the centreline velocity at distance x from the face is roughly

so a hood that holds 0.5 m/s at 300 mm holds a quarter of that — 0.125 m/s — at 600 mm, and a mere 0.055 m/s at 900 mm, where ordinary room air currents (0.1–0.5 m/s near a welder's movement) simply walk away with the plume. Doubling the distance costs 4× the airflow to compensate (for large x, Q \propto x^2). There is no practical flow rate at which a hood three metres from the arc works; the airflow needed grows as the square of an already-lost argument. This is why §14 says as near as possible and enclosed so far as possible — enclosure tries to replace the square law with a simple face area.

Now run the numbers on the alternative strategies for a single shielded-metal-arc welder. Measured fume generation rates for mild-steel SMAW run to 312 mg/min for the arc-on period [7]. Over an eight-hour shift at 30 % arc-on (144 minutes of arc time), that is

Dilution route: while the arc burns, fume enters the room at 312 \times 60 = 18{,}720 mg/h. To hold the breathing zone at even 1 mg/m³ — one-fifth of the OSHA PEL — a perfectly mixed room needs

of clean air, per welder, while the arc is on. (Shift-averaged across the 30 % duty cycle the requirement falls to about 5,600 m³/h — still the airflow of a small spray booth for one welder, and real rooms are never perfectly mixed, so the breathing zone sees worse than the average.) You would be trying to ventilate the welder's air from the wrong end of a 20-metre problem.

Capture route: a 1,240 m³/h extraction arm (computed in §2.3 — that is exactly what the capture-velocity equations produce) at 300 mm captures, in field conditions, of the order of 90–95 % of the fume at generation. Residual emission: 5 % × 18,720 = 936 mg/h, which needs only 936 m³/h of dilution to hold 1 mg/m³ — less than the arm itself already moves. The comparison in air: 18,720 vs 1,240 m³/h — a factor of fifteen. And the exhausted air carrying the fume goes to a filter instead of a breathing zone. This one arithmetic is the reason LEV sits above dilution in every code, guideline and legal sentence ever written on the subject [1][2].

2.2 Capture velocity — the table that sizes everything

The design flow of a hood is set by the capture (control) velocity required at the point of generation — the air speed needed to entrain the contaminant and carry it to the hood against room currents and the plume's own momentum. ACGIH and ASHRAE publish the table; it is short enough to memorise and old enough to trust [2]:

Condition of contaminant dispersion · Examples · Capture velocity

Released with essentially no velocity into still air · Evaporation from tanks, degreasing, plating · 0.25–0.5 m/s (50–100 fpm)

Released at low velocity into moderately still air · Container filling, conveyor transfers, welding · 0.5–1.0 m/s (100–200 fpm)

Active generation into a zone of rapid air motion · Barrel filling, crushing, cool shakeout · 1.0–2.5 m/s (200–500 fpm)

Released at high velocity into a zone of very rapid air motion · Grinding, abrasive blasting, tumbling, hot shakeout · 2.5–10 m/s (500–2,000 fpm)

Note what the table does to two neighbouring booths. A welding arc, diffuse and lazy, is controlled by 0.5–1.0 m/s — achievable with a movable arm 300 mm away. A grinding wheel throws abrasive and metal at high velocity (the brief flight speed of particles from a 35 m/s wheel is tens of m/s), so the required capture velocity is up to 10 m/s and no practical movable hood can do it; grinding is controlled by enclosure — a booth, a curtained cabinet, a downdraft table — where the extraction only needs to win at the face opening (typically 0.5–1.0 m/s through the opening for a well-designed booth). The same 2,000 fpm ceiling appears in the abrasive-blasting guidelines as a simple rule: 500 ft/min (2.5 m/s) inward through every opening of a blasting cabinet or room, or 200 cfm per square foot of open area where curtains replace a solid door [22].

2.3 Worked: the 1,240 m³/h welding arm

Design a fume arm to hold the capture velocity v_c = 0.5 m/s at the arc, with the hood face 300 mm away and a 150 mm round flanged opening. The flanged-opening equation (a circular hood with a flange against a surface wastes far less air than a bare pipe):

with x = 0.30 m, A = \pi (0.075)^2 = 0.0177 m²:

Cross-check against the market: portable fume extractors in the Indian market move 1,750–2,850 m³/h per unit and sell for ₹65,000–85,000 [28]; commercial arms are quoted for 1,000–1,500 m³/h each. The equation and the catalogue agree, which is how you know both are honest. Two refinements worth knowing:

2.4 Dilution ventilation: the limited tool it actually is

Dilution is not useless — it is insufficient alone and appropriate as a supplement: general exhaust to keep background concentrations from accumulating between booths, to sweep fugitive emissions away from adjacent workers, and to manage the gas fraction (CO, ozone, NOx) that no filter catches. The classic hotel-load rule of thumb — supply the greater of the dilution requirement and 10 m³/min per person — is a floor, not a design. Two mechanical details always confound dilution systems:

3. Transport: Ducts and the Velocity That Keeps Dust Dumb

3.1 Transport velocity — the second table

Once a hood has captured the dust, the duct system must keep it moving. Below a material-dependent velocity, particles drop out of suspension (saltation), settle on duct floors, and begin a quiet career of narrowing the duct, unbalancing the system and — if the dust is combustible — stocking a fuel bed. The design rule is always the same: size every duct from a minimum transport velocity, not from a friction rate. Duct diameter is then computed from A = Q/v and rounded down to the next standard size — the opposite of HVAC practice — because a duct one size too large drops below the minimum and the dust elects to stay [23][24]:

Material · Minimum transport velocity · In m/s · In fpm

Gases and vapours · — · 5–10 · 1,000–2,000

Fumes (< 1 µm, low density) · 10.2 · 10–13 · 2,000–2,500

Very fine dust (1–10 µm, light) · 12.7 · 12.7–15.2 · 2,500–3,000

Dry dusts and powders (10–100 µm) · 15.2 · 15.2–17.8 · 3,000–3,500

Average industrial dust · 17.8 · 17.8–20.3 · 3,500–4,000

Heavy dust and chips (metal, >100 µm) · 20.3+ · 20.3–22.9 · 4,000–4,500

Metal turnings, heavy/wet · 22.9 · ~23 · 4,500

Two consequences deserve emphasis before the numbers start. First, transport velocity is a floor, not a target — running a wood-dust duct at 25 m/s instead of 18 wastes fan power as v^3-ish and wears the duct, while running it at 15 m/s builds a sawdust log. Second, there is a design ceiling too: at junctions, the classic rule is to keep the velocity rise through a junction within about 20 %, so elbows and entries are eased, branch angles held to 30–45°, and expansions kept under 60° included angle [24]. Dust-air ducts are, in effect, a conveyor: the air is the belt, and belts do not like speed changes mid-transfer.

3.2 Pressure and velocity: the two-aircraft vocabulary

Every measurement in a duct system is expressed in three pressures, and the entire discipline becomes simple once they are separated:

Conversions worth carrying in your head: 1\ \text{m/s} \approx 197\ \text{fpm}; 1\ \text{in.wg} = 249\ \text{Pa}; standard air \rho = 1.2\ \text{kg/m³} (0.075 lb/ft³). A 14 m/s main carries VP = 0.5 \times 1.2 \times 14^2 \approx 118 Pa ≈ 0.47 in.wg. Every fitting in the system turns a multiple k of that number into heat: a 90° elbow at r/D = 1.5 costs about 0.2–0.3 VP, a plain duct-entry H_e = 1.0 VP, a well-flanged hood entry 0.25–0.5 VP [1].

3.3 The system, worked end to end

Design a central weld-fume extraction system for eight welding stations, each with a 1,240 m³/h arm (just computed), plus ducting to a cartridge filter in the yard and back through a fan to a stack. Design conditions: galvanised duct, \varepsilon = 0.15 mm; air at 20 °C, \rho = 1.2, \mu = 1.81\times10^{-5} Pa·s.

Branch sizing (one per station): Q = 0.344 m³/s at the minimum fume transport velocity 12.5 m/s gives required area 0.0275 m² → d = 187 mm → round down to 180 mm standard duct. Actual velocity:

comfortably above the 10–13 m/s fume-floor and below the wear zone.

Main sizing: eight branches → Q = 8 \times 1{,}240 = 9{,}920 \approx \mathbf{10{,}000\ m³/h} = 2.76 m³/s. At the same 12.5 m/s floor: area 0.220 m² → d = 530 mm → round down to 500 mm. Actual main velocity 14.0 m/s.

Friction, done properly. Duct friction is Darcy–Weisbach: \Delta p = f\,(L/D)\,\tfrac{1}{2}\rho v^2, with f from Colebrook–White (this is the check that keeps the numbers honest — the "friction chart" is just this equation, and galvanised spiral duct is not glass-smooth):

Branch (180 mm, 13.5 m/s): Re = 1.6\times10^5, f = 0.0207 → \Delta p = 0.0207 \times (1/0.18) \times 110 = \mathbf{12.6\ Pa/m}. Main (500 mm, 14.0 m/s): Re = 4.6\times10^5, f = 0.0164 → \Delta p = \mathbf{3.9\ Pa/m}. Note the shape of the problem: the branch is a 3,200 Pa-per-100-m duct and the main is a 390 Pa-per-100-m duct, because friction rises as v^{1.8\text{–}2}/D^{1.2} and dust systems force the velocity up while rounding the diameter down. In dust work, velocity is a requirement and friction is its tax.

The total-pressure budget — the number the fan must produce — assembled over the index run (the longest, worst branch), at the design condition of a dirty filter:

Element · Calculation · Loss

Hood entry (flanged, H_e = 0.5) · 0.5 \times 110 Pa · 55 Pa

Branch duct, 10 m · 10 \times 12.6 · 126 Pa

Branch fittings (2 elbows, k = 0.25 each) · 0.5 \times 110 · 55 Pa

Branch junction into main (k = 0.2) · 0.2 \times 110 · 22 Pa

Main duct, 25 m · 25 \times 3.9 · 97 Pa

Main fittings (3 elbows, k = 0.25) · 0.75 \times 118 · 89 Pa

Cartridge filter, dirty · design \Delta p · 1,000 Pa

Outlet duct to stack, 6 m · 6 \times 3.9 · 23 Pa

Stack exit (k = 1.0 of VP) · 118 · 118 Pa

Subtotal · · 1,584 Pa

Design margin +10 % · · ≈ 1,740 Pa (7.0 in.wg)

Notice the punchline in that table: the filter alone is 63 % of the system's pressure demand. The ductwork that everyone argues about is the small change; the media is the machine. Fan sizing follows directly:

The 0.62 includes a belt-drive and installation allowance — a bare fan at its peak might touch 78 %, but "system effect" (bad inlet boxes, elbows hard against the fan, a discharge that reverses the rotation) reliably eats 10–30 % of delivered flow, which is why the installation details in AMCA's literature deserve as much attention as the catalogue curve [25].

3.4 Details that decide whether the arithmetic survives contact with the shop

4. Air Movement: Fans, Curves and the Cube Law

4.1 The family and its efficiencies

A fan is a device for turning shaft power into air horsepower, P_{air} = Q\,\Delta p — the product that this entire guide has been building toward. The selectable family, with the efficiency bands you will see at the operating point (peak total efficiency, from manufacturers' published curves and the ISO 12759 framework [26]):

Type · Best at · Peak efficiency band · Watch-outs

Centrifugal, backward-curved / aerofoil · Clean to lightly dust-laden air, high pressure · 75–86 % · The default choice for dust systems; robust, efficient, non-overloading

Centrifugal, forward-curved · Low pressure, compact (AHU duty) · 55–70 % · Very sensitive to dust build-up; poor for particulates

Centrifugal, radial / radial-tip · Abrasive/dusty streams, high pressure · 55–75 % · Lower efficiency; the sacrificial choice for grinding dust & shot blast

Mixed flow / vaneaxial · High flow, moderate pressure, inline · 70–85 % · Swirl losses; needs good inlet conditions

Tubeaxial / propeller · Very high flow, tiny pressure · 40–60 % · Dilution ventilation only — cannot feed a ducted system

For an industrial extraction system, the decision tree is short: backward-curved centrifugal for fume and light dust; radial-tip or a wear-lined impeller for abrasive or heavy loading; axial only where the system pressure is trivial. Efficiency is not vanity: at 10,000 m³/h and 1,740 Pa, every efficiency point is 2.76 \times 1{,}740/1000 \times (1/\eta) of shaft power, and 0.60 vs 0.70 is a 1.5 kW permanent difference.

Also worth knowing the boring parts: AMCA 210 / ISO 5801 define the test rig (a fan's curve means nothing without the measurement standard behind it), ISO 12759 defines efficiency classes (FEG — fan efficiency grade — from ~20 to ~85 in points), and the EU's fan regulation 327/2011 turns that into minimum-efficiency law for 125 W–500 kW machines in its jurisdiction [26]. India has no equivalent hard regulation yet; which is precisely why specifying an efficiency grade and demanding the tested curve in writing is on you.

4.2 The fan curve and the system curve — where the system actually operates

A fan's catalogue shows pressure as a decreasing function of flow (a rising-then-falling curve for backward-curved; stall-prone at the left edge). The ductwork imposes system resistance rising with the square of flow: \Delta p_{sys} = \Delta p_{static} + k Q^2. The operating point is the intersection of the two curves — and nothing the fan's nameplate says survives that intersection:

4.3 The affinity laws — the most valuable equations in this guide

For any centrifugal fan at constant system geometry, changing speed changes everything in powers of two and three:

Flow scales linearly, pressure with the square, shaft power with the cube [27][28]. The cube is where the money lives:

Speed · Flow · Pressure · Power · Power saved

100 % · 100 % · 100 % · 100 % · —

90 % · 90 % · 81 % · 72.9 % · 27 %

80 % · 80 % · 64 % · 51.2 % · 49 %

62.5 % · 62.5 % · 39 % · 24.4 % · 76 %

50 % · 50 % · 25 % · 12.5 % · 87.5 %

The US Department of Energy's worked example makes it concrete: an exhaust fan drawing 16.4 kW at full speed needs only 20\ \text{hp} \times (1/2)^3 = 2.5 shaft hp at half speed, and even after real motor (77.8 %) and drive (86 %) efficiencies, the input power is 2.8 kW — an 82.9 % reduction [27].

Two honest caveats, because the cube law is oversold in brochures. First, it assumes the system has no static-pressure component; real ducted systems sit between exponents 2.0 and 3.0, with field-measured values commonly 2.3–2.6 for pressure-controlled systems [29]. Second, the cube law does not apply when you keep the speed constant and throttle with dampers: a damper steals pressure without giving the fan the speed reduction that is the actual prize, which is why "we close the blast gates" saves ~10 % and "we added a VFD" saves ~50+ %. The full comparison, worked on the 10,000 m³/h system from §3:

Legacy: fixed-speed fan, blast gates, 6,000 operating hours/year (two shifts, 300 days). The fan grinds at effectively constant power (gates shift the operating point toward slightly lower flow and power, but the motor still eats ≈ 90 % of rated input, ≈ 8.0 kW):

Upgraded: VFD with duct-pressure control (sensor at two-thirds of the main run; pressure setpoint trimmed so the fan makes only the pressure the system needs), with the realistic exponent 2.5 and a usage profile: 1,200 h at all eight stations (100 % flow), 3,000 h at five stations (62.5 %), 1,800 h at two-to-three stations (35 %):

Mode · Hours · Speed · Power · Energy

All stations · 1,200 · 100 % · 8.9 kW · 10,700 kWh

Five stations · 3,000 · 62.5 % · 8.9 × 0.625^2.5 = 2.7 kW · 8,200 kWh

Two-three stations · 1,800 · 35 % · 8.9 × 0.35^2.5 = 0.7 kW · 1,200 kWh

VFD total · · · · ≈ 20,000 kWh/yr

The swing: 48,000 → 20,000 kWh — 28,000 kWh saved, a 58 % cut, ₹2.24 lakh a year at ₹8/kWh of industrial power, which pays for a 11 kW VFD within a single financial quarter, and then keeps paying. (The ideal-cube numbers would be even better — 0.625³ = 24.4 % instead of 30.9 % — but this guide prefers the defensible version.) Do not skip the engineering details that make it safe: a minimum speed to keep duct transport velocity above the settling floor and filter face velocity out of the re-entrainment zone, ramped acceleration, and the motor's thermal duty at reduced cooling. A VFD on a dust system is not a dimmer switch; it is a second control system, and it is worth every rupee of the extra complexity.

5. Collection I — Cyclones: Big Force, Serious Limits

5.1 The physics in one paragraph

A cyclone converts velocity into swirl; swirl centrifuges particles outward (a 2 g-to-2,000 g acceleration depending on diameter and inlet speed); the wall drags them down into a cone that delivers them — hopefully — into a hopper while cleaned air leaves through a vortex finder. No media, no cleaning cycle, no consumables: the cyclone is the only dust collector that never clogs, because it has nothing to clog. Its weakness is equally structural: the separation force is weakest exactly where the dust is finest, and the particles you care most about (submicron fume, the respirable cut) have almost no mass to centrifuge.

5.2 The canonical geometry and its numbers

The classical high-efficiency proportions are named for Stairmand (1951), and one family recurs across every retrofit in the world: the 1D3D (1 diameter of cylinder, 3 diameters total height) and 2D2D geometries, whose design ratios let you dimension a cyclone from its diameter alone:

Dimension (1D3D, Stairmand HE) · Ratio · 750 mm example

Body diameter D · 1 · 750 mm

Inlet height a · D/2 · 375 mm

Inlet width b · D/4 · 186 mm

Cylinder height h · 1D · 750 mm

Total height H · 3D · 2,250 mm

Outlet (vortex finder) D_e · 0.5–0.6 D · ~450 mm

Dust outlet · 0.25–0.375 D · ~190–280 mm

Worked — a wood shop's cyclone at 4,000 m³/h. Inlet velocity designs run 15–18 m/s (below that, mediocre separation; above, pressure drop and abrasion with no efficiency dividend worth the fan power). At v_i = 16 m/s: inlet area A = 4{,}000/3{,}600/16 = 0.0694 m²; with A = a\,b = D^2/8, we get D = \sqrt{8A} = 0.745 m → 750 mm. The inlet then measures 186 × 375 mm, the cylinder 750 mm tall, the cone another 1,500 mm, dust exits through ~190 mm, and the whole thing stands about 2.5 m plus hopper and standpipe.

Pressure drop. Cyclone loss is most usefully expressed in velocity heads of the inlet: \Delta p = \Delta H \cdot \tfrac{1}{2}\rho v_i^2, where \Delta H (a function of geometry only) runs around 8–9 inlet velocity heads for classic designs. At 16 m/s, VP_i = 154 Pa and \Delta H \approx 8 gives ≈ 1,200 Pa — which lands right on the measured value for a 1D3D at 16 m/s (3,200 fpm): 4.5 in.wg ≈ 1,120 Pa [13]. And a beautiful property falls out of the matching exercise: within a fixed geometry family, cyclone pressure drop is approximately independent of size — a 750 mm and a 1,500 mm Stairmand cyclone cost the same static pressure at the same inlet velocity, which is why you cannot buy your way out of pressure drop with a bigger cyclone; you buy less efficiency with it.

Cut point. The classic Lapple model d_{50} = \sqrt{\,9\mu b\,/\,(2\pi N v_i \rho_p)} ties the 50 %-capture particle size to geometry and conditions. For our 750 mm unit with b = 0.186 m, effective turns N \approx 6, v_i = 16 m/s and wood-dust particle density 650 kg/m³:

Read that number with care, because it is the whole story of cyclone mis-selling: half of all 8.8 µm particles escape, and essentially everything smaller does. If your dust has a mass-median diameter of 20 µm (typical sanding dust), the cyclone's mass efficiency looks respectable — 85–90 % — because the mass lives in the coarse tail. But 90 %+ of the particle count — and the entire respirable fraction below 5 µm that deposits in lungs — passes straight through, and roughly 15–25 % of the emitted mass sails out the stack of a well-maintained cyclone on fine dust, more when it is overloaded or the hopper is leaking. That is why every serious regulatory guide classifies cyclones as pre-cleaners: they knock down the coarse load, protect the real filter from 80–90 % of the solids, act as a spark arrestor and flame front disrupter for hot processes, and never as the last stage before a breathing zone. The Indian market prices this honesty at ₹35,000 for a basic unit to ₹1–2 lakh for a multi-cyclone assembly — the cheapest coarse separation money can buy [28].

Where cyclones earn their place: wood shops (the dust is coarse and the fire load matters), grinding rooms ahead of cartridges, shot-blast exhausts, foundry sand handling ahead of baghouses, any hot or spark-laden process, and as the first stage of a two-stage system where the budget cannot yet afford Stage Two — in which case write down that the stack is still emitting the fine fraction, and put it on the capital plan.

6. Collection II — Fabric and Cartridge Filters

6.1 The governing variable: air-to-cloth ratio

A fabric collector's whole design reduces to one number: the air-to-cloth ratio (filtration velocity), the airflow divided by the total filter media area, expressed as m/min or ft/min. The physics of the choice: the dust cake does the real filtering, so the design must let the cake form and release at a rate the pulse-cleaning system can sustain. Run too fast and the cake compresses into the media (blinding), emission and pressure drop climb, and bag life collapses; run too slow and you have bought steel and media you will never use. Representative design bands, by cleaning mechanism and dust [15][16]:

Collector type / dust · Air-to-cloth ratio

Shaker-cleaned fabric · 0.6–1.0 m/min (2–3 ft/min)

Reverse-air fabric (fine, cohesive dusts) · 0.45–0.75 m/min (1.5–2.5)

Pulse-jet fabric, general dust · 1.0–1.5 m/min (3.3–5)

Pulse-jet, coarse wood dust · up to 1.5–1.8 m/min (5–6)

Pulse-jet, fine metal fume / weld fume · 0.6–1.0 m/min (2–3.3)

Cartridge collectors (pleated media, higher media density) · 0.3–0.6 m/min (1–2) effective

Two field notes from real plant data. First, the ratio is a correctable mistake with measurable dividends: lowering a cement plant's A/C from 6.2 to 4.1 ft/min raised bag life 34 % and cut pressure drop 28 % — about $47,000/yr in maintenance [15]. Second, "gross" vs "net" counts: pulse-jet cleaning takes rows off-line for a fraction of a second, and bags blanked or broken shrink the effective area; losing 15 % of the cloth area quietly raises every remaining bag's duty by 18 % [16]. The differential-pressure gauge is what tells you this has happened — which is why it should be installed, logged, and believed.

The companion variable is can velocity — the upward air speed in the open space between the bags — which must stay below about 2.5 ft/s (0.75 m/s) in pulse-jet units, and below 2.0 ft/s (0.6 m/s) for dusts finer than 10 µm, or freshly pulsed dust falls back onto its colleagues instead of into the hopper (re-entrainment) [15]. This is the number that sizes the housing after you have sized the media, and it is the number that a supplier quoting by price alone will have quietly violated.

6.2 Worked: the cartridge collector for the 10,000 m³/h weld system

Weld fume is fine (dominant mode 0.22 µm), sticky, and in demand of conservative filtration: design at v_f = 0.5 m/min (1.6 ft/min). Required net media area:

With standard 22 m² pleated cartridges (\sim325 mm diameter × 660 mm), that is 16 cartridges — a two-tier, walk-in-sized housing you have seen a hundred times outside a fabrication shop without knowing its arithmetic. The design table for sensitivity:

Filtration velocity · Media area · Cartridges

0.4 m/min (1.3 ft/min) · 417 m² · 19

0.5 m/min (1.6 ft/min) · 333 m² · 16

0.6 m/min (2.0 ft/min) · 278 m² · 13

Weld fume purchases its cloth like a pessimist on purpose: the difference between 13 and 16 cartridges is roughly ₹1–1.5 lakh of media — about one year of not buying replacement filters because they blinded in eighteen months instead of four years.

Baghouse equivalent for a larger, coarser duty (wood dust, 20,000 m³/h, pulse-jet at 1.2 m/min): 278 m² of cloth; with 130 mm × 2,000 mm bags (0.82 m² each) that is ≈ 340 bags, plus can velocity and hopper design (60–70° hopper angles), and a walk-in plenum because someone is going to have to change those bags.

6.3 Pulse-jet mechanics and the media menu

The pulse-jet — a blast of compressed air (80–100 psi through a venturi) that snaps each bag against its cage to shed its cake — is more than 70 % of new installations worldwide [15]. Its maintenance-critical parameters are few, and worth writing on the control panel: pulse pressure ~90 psi, pulse duration 100–150 ms, and pulse interval 8–12 s — over-pulsing (intervals below ~5 s) is measured to cost up to 40 % of bag wear, under-pulsing drives pressure drop up. Solenoid valve quality (the Goyen-class workhorses) decides whether this cadence survives 10⁷ cycles.

Media selection is chemistry first, temperature second, price third:

Media · Continuous temperature · Best for · Avoid

Polyester felt · 130 °C · The default for wood, most metal dusts, weld fume · Hydrolysis above 40 % RH + heat; sparks

Polypropylene · 90 °C · Acid mists, chemicals · Heat

Aramid (Nomex) · 190 °C · Asphalt, foundry, cement, hot processes · Hydrolysis with moisture + acid

PTFE (Teflon fabric) · 250 °C · Aggressive chemicals; hot gas · Cost (5–10× polyester)

P84 / PI blends · 240 °C · Fine dust capture, high temp · Cost

Antistatic (carbon-loaded, earthed) · as base · Combustible metal & polymer dusts · — (mandatory, not optional, for St-class dusts)

PTFE membrane on any felt · as base · Weld fume, sticky submicron dusts (surface filtration, easy release) · —

For welding fume, that table's answer is specific: polyester with a PTFE membrane (H class media keeping emission below ~0.5–1 mg/m³), antistatic construction if metal powder is present, and a secondary HEPA stage if the air is to be recirculated indoors. Which brings the most valuable single number in weld-shop ventilation:

Recirculation is what lets a weld shop afford filtration. HEPA-grade filtration at 99.97 % at 0.3 µm (EN 1822 H13/H14) turns 10,000 m³/h of fume-laden exhaust into indoor-quality supply, and the shop stops paying to heat, cool and replace that air 6,000 hours a year. The F9-pre-filter + H14-HEPA stack, monitored by a differential-pressure gauge that triggers filter change, converts §2's make-up-air bill (50 kW of heating in winter) into a rounding error. The catch: recirculation quality is only as good as the leak you cannot see, so it demands the differential gauges and a smoke-test commissioning that a throwaway install never gets.

6.4 What the filter costs in electricity (the invisible bill)

The same table that gave the filter 63 % of the system's pressure demand gives it a permanent energy bill: at 2.76 m³/s through a 1,000 Pa filter, the media alone dissipates

And the derivative: each +250 Pa of avoidable filter pressure drop pays ₹33,000 a year to have been ignored. The gauge with the red needle costs ₹3,000; the failure to read it costs a cartridge set and a third of a lakh of electricity annually.

6.5 The rest of the zoo, briefly

7. The Explosion: Combustible Dust Is a Different Discipline

7.1 The pentagon, and why every shop is closer to it than it thinks

A dust explosion needs five things simultaneously — the dust explosion pentagon: combustible dust, oxygen, an ignition source, dispersion (a cloud), and confinement. Remove one and you have a fire, a puff, or nothing. The classic disasters removed none; the reason this section exists is that ordinary fabrication processes keep assembling all five by accident:

Dust explosion class · Kst (bar·m/s) · Explosion severity · Representative dusts

St 0 · 0 · Non-explosible · Limestone, sand, cement (verify by test; do not assume)

St 1 · 1–200 · Weak to moderate · Wood 100–200, coal 55–200, sugar 75–160, most plastics, grain, PVC

St 2 · 201–300 · Strong · Cellulose ~200–230, cornstarch can reach this band, some polymers/pigments

St 3 · >300 · Very strong · Aluminium fine (<75 µm): 400–700; magnesium 500+; titanium ~412

The parameter in that table is measured, not guessed: a 20-litre or 1 m³ test vessel per ISO 6184-1/EN 14034/ASTM E1226 disperses the dust, ignites it with a 10 kJ chemical igniter, records the pressure rise, and applies the cubic law to normalise for vessel size [19]:

Companion parameters: Pmax (peak explosion pressure — 7–10 bar for organics, 10–15 bar for metals), MIE (minimum ignition energy — under 1 mJ for fine aluminium against 40–60 mJ for cornstarch and 60–100 mJ for coal; a person's static discharge is 10–30 mJ, hence aluminium's reputation), MEC (minimum explosible concentration, typically in the tens of grams per m³ for organics — concentrations that occur inside collectors and duct but not in room air), and layer ignition temperature, which is about settled dust on hot surfaces.

7.2 The arithmetic that separates wood from titanium

Here is the sentence to take away: the same collector needs a different explosion-protection design for every dust class it might hold, and the difference is a factor of three or more. Per NFPA 68's vent-sizing framework, the required vent area scales approximately as

where V is the enclosure volume and P_{red} its reduced (survivable) explosion pressure. Same volume, same vent strength: an aluminium dust at K_{st} = 450 needs roughly 3.2× the vent area of a wood dust at 140 bar·m/s [19] — and in fact the honest statement in the modern guidance is stronger: for St 3 metal dusts, passive venting alone may be insufficient regardless of area, and engineered isolation and suppression become the discussion [19]. A vent-sized-for-grain dust collector handling aluminium is not under-protected; it is a bomb with a decorative hatch.

7.3 The rulebook: NFPA 660 and the Dust Hazard Analysis

American practice for a century arrived as a stack of commodity standards — NFPA 652 (fundamentals), 61 (food), 484 (metals), 654 (general/plastics/chemicals), 655 (sulfur), 664 (wood) — and in 2025 the NFPA collapsed all six into a single NFPA 660, Standard for Combustible Dusts and Particulate Solids: fundamentals in Chapters 1–10, commodity chapters after (wood is Chapter 14, metals 12, general 13/15) [17]. The consolidation matters less than its centrepiece, which carried over intact: the Dust Hazard Analysis (DHA) — a documented, methodical review of every process, vessel, duct and building compartment that handles combustible particulate, identifying where the pentagon assembles and what protects it. The 2020 edition of the predecessor standard also folded in explicit additive-manufacturing and ultrafine-powder provisions — a direct acknowledgement that printable metal powder is now handled outside foundry walls [18]. And the second centrepiece is housekeeping, because the layer is the fuel supply for the secondary explosion: the long-standing criterion in the general standard is that a dust layer of 1/32 inch (0.8 mm) thickness over 5 % of floor area — and equivalent layer depths on beams, ledges and ducts — converts a survivable primary event into a building-clearing secondary one, and must trigger corrective action [18]. NFPA 654's historical text is blunt that the requirement is met by cleaning programs run on a frequency the facility can prove, with vacuum (bonded/conductive equipment), never compressed-air blow-down that disperses exactly the cloud the standard exists to prevent.

The protection menu, in the order cost-efficiency usually puts it:

  1. Prevent — substitution where possible (fewer fines), housekeeping, ignition-source control: bonding & grounding everywhere, no unclassified electrics in classified zones, hot-work permitting, spark detection/quench systems on duct inlets from hot processes.
  2. Protect the enclosures — explosion venting (NFPA 68: vent panels/doors, flameless vents where the vent line cannot be near occupied space, vent ducts sized for the Kst), explosion suppression (NFPA 69 chemical isolation — the high-end answer for indoor or hard-to-vent equipment), and isolation to stop flame propagation between vessels: rotary airlocks, double-flap valves, fast guillotine valves, material chokes.
  3. Assume some leak — design the room: pressure-relief paths, fire separation, and no occupied space on the vent axis.

For the small Indian shop, the actionable core of all of the above compresses to five items: (i) identify whether any of your dusts are combustible (if wood/aluminium/titanium/sugar/fine plastic — they are; a screening test is cheap), (ii) get a DHA-grade look — even a competent internal review by someone who knows NFPA 660 — at every collector, duct, cyclone, hopper and dust-producing process, (iii) ground and bond the duct system and collector, (iv) run the housekeeping program with the 1/32-inch criterion as the trigger, and (v) never, ever site a dust collector indoors without explosion-rated design — a leading share of the casualties in dust disasters are the collector's own panels finding their way into the building.

8. The People: Exposure Arithmetic That Decides the Design

8.1 Fume generation rates — what the arc actually emits

The air requirements of welding fume are set by fume generation rate, measured in the laboratory with the chamber and collection methods of the standard studies. The spread across processes is enormous — a factor of nearly 10× between the worst and best common arcs, on mild steel [7][8]:

Process (mild steel unless noted) · Fume generation · Normalised (per g of electrode) · Notes

SMAW (stick) · 312 mg/min · 12.6 mg/g · The dirtiest common process; also the most used (45 % of consumable usage)

Globular GMAW (CO₂) · 387 mg/min · 6.4 mg/g · High, but fast travel partially offsets

FCAW · 285 mg/min · 4.7 mg/g · Middle

Spray GMAW · 212 mg/min · 3.5 mg/g · —

Short-circuit GMAW · 58–113 mg/min · 2.3–4.5 mg/g · The workhorse substitute for stick

STT / RMD · 84–88 mg/min · 3.3–3.5 mg/g · —

Pulsed-spray GMAW · 89 mg/min · 1.5 mg/g · High-productivity, low fume

CMT · 35 mg/min · 0.92 mg/g · The low-fume champion

Stainless SMAW (E308) · 228 mg/min · 8 mg/g · And Cr(VI) 7,800 µg/min

Stainless pulsed-spray · 15 mg/min · 0.2 mg/g · Cr(VI) 50–130 µg/min — a 150:1 improvement

Read the last two lines together: switching stainless work from stick to pulsed-spray GMAW cuts total fume by 15× and hexavalent chromium by over a hundred×, at lower cost per metre of weld because time is the dominant term. Process substitution is the strongest engineering control in the entire welding chapter — stronger than any fan. The rest of the hierarchy, in effectiveness order: on-torch extraction (90–95 % capture at the arc) → local exhaust arms → general ventilation → respirators (the legal last resort, and an administrative failure if it is the primary control). The standards documents say the same thing in five languages and 80 pages [3][4].

8.2 Why the total-fume number protects the wrong thing

Return to the worked shift: 44.9 g of fume, and the ladder of what it is made of. Manganese in welding fume runs roughly 0.3–1.5 % of fume mass for common mild-steel consumables (upward of that for some wire/flux chemistries); Cr(VI) for stainless during SMAW converts at rates measured around 55 % of the chromium (versus ~5 % for GMAW). Now overlay the limits:

of clean air per welder while the arc burns — a hundred times the flow of a fume arm, and still assuming perfect mixing. Shift-averaging at 30 % arc-on only brings it down to ~28,000 m³/h. There is no ventilation strategy at this scale; the only strategies are substitution (low-Cr(VI) process), extraction at the arc with HEPA-grade final filtration, and personal protection for the residual. The microgram limit is the reason "fume extraction" for stainless is not a comfort amenity.

8.3 Silica, the other mass killer, and India's specific problem

Respirable crystalline silica — concrete cutting, stone working, foundry sand, blasting substitute — has a limit structure that India has been unsuccessful in defending on. The numbers: OSHA PEL 0.05 mg/m³ (2016 rule), ACGIH TLV 0.025 mg/m³, India's DGMS limit for mines 0.15 mg/m³ — three to six times looser — and yet measured silicosis prevalence in Indian stone-mining regions runs 12–30 %, including in cohorts whose measured exposures (mean RCS 0.12–0.17 mg/m³ in sandstone, masonry and granite mines) sat near or below even the loose national limit [21]. The lesson generalises beyond mining: when exposures hover at the limit and disease is endemic, the limit is the problem — design to the strictest credible number, not the local legal one. For shop-floor silica work (stone countertop cutting, concrete grooving, sand blasting), that means wet suppression plus HEPA-class LEV at the tool, measured by respirable sampling, treated like the carcinogen it is.

8.4 The measurement layer — what to instrument, and why

You cannot manage this section on belief. The minimum viable hygiene instrument set for a fab shop:

9. Design, Commissioning, Maintenance: Making the Arithmetic Survive

9.1 The specification checklist (in the order of leverage)

  1. Substitute first. Whatever the process, ask what reduction is available before the fan: low-fume welding modes, wet cutting for silica, enclosed blasting, powder handling in glovebox-grade enclosures.
  2. Capture at the source, enclosed if possible. Every hood gets its flow from §2's equations, sized for the nearest practical position, not the furthest dream.
  3. Size ducts by transport velocity (§3), round down, count fittings honestly (every elbow is a tax you can measure), and keep junction velocity changes small.
  4. Select collector media by dust chemistry and size, at a conservative air-to-cloth ratio, with the differential gauge in the specification, not the options.
  5. Fit the fan to the dirty-filter operating point (§4), specify efficiency grade and demand the tested curve; plan the VFD from day one — the sensor, the harness, the minimum-speed settings (the drive itself can be added later, but the motor and the controls case cannot be retrofitted cheaply).
  6. Plan the air balance — make-up air, recirculation vs exhaust, pressure relationships between rooms — before the first duct hanger goes in.
  7. Write the DHA and the housekeeping programme into the same document as the equipment list (§7) if any dust is combustible — because that is how they will actually be executed.
  8. Instrument and schedule — gauges on filters, anemometer rounds, sampling calendar, filter-change stock (filters are consumables; budgeting for them once is cheaper than discovering it in a failure).

9.2 Commissioning by measurement

A ventilation system is commissioned when, and only when, the following are recorded in writing: total flow at each hood (face velocity × area, or duct traverse); capture velocity at the source distance (smoke or anemometer); fan speed and amps; filter pressure drop clean; make-up air flow; and room pressure. That document is the baseline every future audit checks against, and it costs one shift of a technician's time. Systems without it drift silently: the re-arranged booth that lost its capture, the duct run that collected its own lint, the filter caked to twice its design pressure (which the fan converts into 20 % less air for the same electricity).

9.3 The twelve-month drift (a true shape of failure)

A small tale from the pattern file, illustrative of hundreds: a weld bay's six arms, commissioned in January at 1,240 m³/h each, measured in December at 780–950. Amps were "normal", so nobody looked. Two findings: four hood dampers had been re-adjusted during a summer re-organisation (unbalanced since), and the cartridge collector's differential gauge — installed, but never read — had been pegged; media pressure drop was 1,850 Pa against the 1,000 Pa design. Cleaning the media cadence, re-balancing to the commissioning sheet, and replacing one clamp-band worth of damaged cartridges took an afternoon and restored ~92 % of design flow. The fan, running the whole year, had been delivering 20 % less air and 2.9 kW more power than it needed. The differential gauge read ₹0; reading it was worth ₹26,000/yr in electricity and the difference between compliance and a story.

10. The Ledger: 2026 Rupee Costs, Energy, and What ₹1 Lakh Buys First

10.1 Capital, from the Indian market

Item · Range (₹, ex-GST) · Notes / source tier

Fume extraction arm, 2–3 m, 160 mm · 23,000–50,000 · Kemper/Bomaksan-class, market listings [28]

Portable welding fume extractor, 2 m arm, ~1,750–2,850 m³/h, HEPA-class · 40,000–85,000 · Per unit; the per-station entry point [28]

Cyclone dust collector, basic to industrial multi-cyclone · 35,000–200,000 · ₹35k basic Coimbatore-built; multi-cyclone units to ₹2L [28]

Silo-top filter unit · 56,000–59,000 · [28]

Cartridge filter elements (159 × 1,000 mm pleated) · ~1,500 each · Consumable; a 16-cartridge set ≈ ₹24k [28]

Bag-filter dust extractor, ~6,000 m³/h (woodworking class) · 1,40,000–2,65,000 · Indian-built units [28]

Baghouse filter unit, industrial · ~4,50,000+ · Mid-size unit listing; scales with cloth area [28]

Welding fume extraction system, 5 HP, 2,850 m³/h, 20 m² media · ~85,000 · Package with collector + arm [28]

VFD, 7.5–11 kW, IP54 · 30,000–60,000 · Market band; the single best energy rupee in the building

The build-up for our eight-station system, from sourced components plus ducting and installation: roughly ₹6–9 lakh as a fully portable build (8 × ₹65–85k units) or a central package (8 arms ~₹2.8L + collector + 11 kW fan + ~60 m of ducted system + installation) — quotes from Coimbatore/Pune fabricators for the central variant land in the same order of magnitude, and which is better depends on cartesian geometry, not taste: portable units win in dispersed bays with occasional welding; central systems win where stations are dense and fume duty is continuous (and central systems can recirculate filtered air, while portables discharge into the room by design).

10.2 Operating cost, per year

Line item · Basis · ₹/yr

Fan energy, fixed-speed (legacy) · 48,000 kWh @ ₹8 · 3,84,000

Fan energy, VFD (same work) · 20,000 kWh @ ₹8 · 1,60,000

VFD saving · 28,000 kWh · −2,24,000

Filter energy (the media's own toll) · 2.76 kW × 6,000 h · 1,33,000

Cartridge media (4-yr life, 16 pcs + labour) · ~₹24k + labour / 4 yr · ~10,000

Make-up air conditioning (if exhausting outdoors) · 50 kW thermal, seasonal · 65,000–4,00,000

Baseline sampling & gauge discipline · 2 rounds/yr · 40,000–80,000

The two decisions that dominate everything: the VFD (₹2.2L/yr) and recirculation vs exhaust (₹1–3L/yr in conditioning). Both are invisible on a quotation that lists only "dust collector — ₹85,000". The procurement lesson of this entire series applies here in its purest form: on air systems, the purchase price is the small number.

10.3 Market context, and the shape of the opportunity

The global industrial dust-collector market is roughly USD 9.8 bn in 2025, growing 5.1 % to about USD 12.5 bn by 2030, with Asia-Pacific the largest region and regulatory tightening the primary driver [30]; the wider dust-control systems market (including wet suppression) runs about USD 21.9 bn, growing ~6.2 %, and the wet-collector segment — the safest answer for explosive dusts — is growing fastest at 6.3 % [30]. Put another way: the world is discovering, shop by shop and inspector by inspector, that the air is a machine — and the runs are on the board for the shops that realise it before the ones that are told.

10.4 What ₹1 lakh buys, in order

  1. *₹20,000 — a differential-pressure gauge and a manometer, installed and read.* (Leverage: tells you when to act on everything else.)
  2. ₹25,000–85,000 — one portable extractor or one retrofit arm for the dirtiest station first (the welding bay with stainless or heavy manganese duty, or the grinding booth).
  3. ₹35,000–60,000 — a VFD on the existing collector, if the motor is inverter duty / can be made so, plus the transport-velocity minimum-speed setting.
  4. Remaining — capture-velocity audit (anemometer and a shift), re-balancing, and sealing up the duct leaks and skylight this guide's §9 drift story hides.
  5. Next lakh, in order: enclosure/downdraft for the grinding booth; HEPA recirculation kit for winter; DHA-grade review if any St-class dust exists.

None of these five items requires a new collector, a consultant's master plan, or a shutdown. Every one of them changes a number in this guide that is currently being paid by default.

Frequently Asked Questions

How much airflow does a welding fume extraction arm need?

Size it from capture velocity, not from a catalogue's "up to" number. Set v_c = 0.5 m/s at the arc (ACGIH's 0.5–1.0 m/s band for low-velocity welding), hood face at 300 mm: Q = 0.75 \times 0.5 \times (10 \times 0.3^2 + 0.0177) \approx 1,240 m³/h per station. Halving the arm-to-arc distance to 150 mm drops the required flow to ~470 m³/h — three times less air, energy and filter, forever.

Why do dust ducts round down to the next size?

Because sizing driver is minimum transport velocity — the speed below which particles settle out (10–13 m/s fume, 18 m/s wood, 20–23 m/s heavy metal). Compute A = Q/v, get a diameter, and choose the standard size at or below it so the actual velocity stays above the floor. Rounding up — the natural instinct from HVAC — drops velocity below the settling threshold and builds a fuel bed in your duct.

Does a VFD really halve a dust collector fan's energy bill?

For a speed-controlled centrifugal fan, yes — near it: power follows the cube of speed (0.8^3 = 51\% at 80 % speed; 12.5 % at half speed), and the DoE's worked example lands an 82.9 % reduction at half flow including real drive/motor losses. Two caveats keep it honest: systems with a static-pressure component run at exponents 2.3–2.6 rather than 3, and damper throttling does not deliver cube-law savings — you need actual speed reduction. A worked 6,000-hour shop system: 48,000 kWh (fixed-speed) → 20,000 kWh (VFD) = ₹2.2 lakh a year saved.

How do I size a dust collector for welding fume?

By air-to-cloth ratio, conservatively. For 10,000 m³/h of weld fume at 0.5 m/min filtration velocity, media area = 166.7/0.5 = 333 m² ≈ 16 cartridges of 22 m². Use PTFE-membrane polyester media, antistatic if metal powder is present, and HEPA (H13/H14, 99.97 % @ 0.3 µm) if recirculating indoors. Weld fume's dominant particle mode is 0.22 µm — this is a submicron-filtration problem, which is why a cyclone alone is not a solution.

Are welding fumes really carcinogenic?

Yes — IARC classified welding fumes as Group 1, carcinogenic to humans, in 2017 (Volume 118), with sufficient evidence for lung cancer; UV radiation from welding is also Group 1 (ocular melanoma). About 11 million people work as welders worldwide and ~110 million more are occupationally exposed. The practical implication: engineer controls as if the fume were the carcinogen it is — substitution, on-torch capture, HEPA finish — rather than treating exposure as a comfort issue.

What's the difference between capture velocity and transport velocity?

Capture velocity happens at the source: it's the air speed needed to pull the contaminant into the hood (0.5–1 m/s for welding, 2.5–10 m/s for grinding) and it sizes the flow. Transport velocity happens inside the duct: it's the speed needed to keep captured dust airborne (10–23 m/s depending on material) and it sizes the diameter. The two live in different equations and both must be satisfied; a system can pass the hood test and still settle its dust in the duct — or keep dust moving in a duct that never captures it.

Which dusts explode, and what do I do about it?

Any dust that burns can explode if fine, dry and dispersed: wood (Kst 100–200), coal, sugar, most plastics — and the serious ones, aluminium (400–700+), magnesium (500+) and titanium (~412), which are St 3 and can be ignited by a static spark from a person (< 1 mJ MIE). What to do: identify and test your dusts; run a Dust Hazard Analysis (NFPA 660, the 2025 consolidation); bond and ground ducting and collectors; never vent explosions into occupied space; use explosion venting/suppression/isolation designed for your Kst and Pmax; and keep layers below the 1/32-inch (0.8 mm) trigger with bonded vacuum equipment — never compressed air.

The Discipline in One Page

Industrial ventilation is one idea — move the air with arithmetic instead of hope — worked through a dozen equations. Capture velocity sizes the hood: 0.5 m/s at the weld arc is a 1,240 m³/h arm, and every doubling of distance quadruples the airflow you will never get back (§2). Transport velocity sizes the duct: 10–23 m/s depending on what must keep flying, rounded down, because a dust duct that is one size generous is a duct that settles (§3). Darcy–Weisbach and its fittings assemble a full ten-thousand-cubic-metre system into a 1,742 Pa pressure budget in which the filter — not the ducting everyone argues about — is 63 % of the bill (§3). The fan meets the system at the intersection of two curves, and the affinity laws make speed the most valuable control: 80 % speed is 51 % power, half speed is 12.5 %, and the shift from a fixed-speed fan to a VFD on a real shop system is 48,000 → 20,000 kWh, ₹2.24 lakh a year (§4). Collection is sized by physics that vendors hide in plain sight: a 750 mm cyclone with a cut diameter near 8.8 µm is an excellent pre-cleaner and a poor final filter; 333 m² of cartridge media and sixteen cartridges hold 10,000 m³/h of submicron fume; the media's own pressure drop pays ₹1.33 lakh a year in electricity (§5–6). The two disciplines underneath are harder and more important: the explosion — where Kst 140 wood and Kst 450 aluminium demand different protection arithmetic from the same collector, NFPA 660 now writes it down in one document, and the 1/32-inch layer is the fuel supply for the secondary event (§7) — and the people — welding fume carries an IARC Group 1 label, the total-fume limit under-protects against manganese and chromium by design gap, Cr(VI) makes dilution ventilation arithmetically impossible for stainless, and India's silica story says design to the strictest credible limit, not the loosest local one (§8). Then the ledger, which is where every number above becomes a decision: ₹2.2 lakh a year waiting in a VFD, ₹1.33 lakh a year hidden in an unread pressure gauge, and the first ₹1 lakh of any shop's air budget going to gauges, one arm, one drive, and an afternoon of measurement — no new machines, no shutdown, just the arithmetic (§10).

That is the machine nobody quoted and everybody owns. The shops that understand it will weld, grind and powder-handle for decades longer, at lower cost, with crews that breathe — and they will have done it with a table of velocities, three equations, and the discipline to read a gauge. The air was always in the budget. This guide just itemised it.

Previous guides in this series: Lean Manufacturing & Production Systems · Belt Conveyors & Bulk Material Handling · Cranes, Hoists & Rigging · Cleanroom Engineering & Contamination Control · Industrial Refrigeration & Cold Chain Systems · Industrial Electrical Power Distribution · Metrology & Dimensional Inspection · Industrial Furnaces, Kilns & Refractories · Structural Steel Design & Fabrication · Pressure Vessels & Storage Tanks · Process Piping & Pipe Fabrication · Industrial Steam Boilers & Steam Systems · Non-Destructive Testing.

[1] ACGIH, Industrial Ventilation: A Manual of Recommended Practice — capture velocity concept and definition, hood entry losses (H_e), minimum design duct velocities, slot/plenum velocity guidance, pressure vocabulary (SP/VP/TP), and the design philosophy of local exhaust at the source. [2] ASHRAE Handbook, HVAC Applications, Ch. 32/33 "Industrial Local Exhaust" (and SI edition Ch. 33) — capture (control) velocity table (0.25–10 m/s across dispersion conditions), Q_o = V_o A_o hood flow relation, slot velocity ~2,000 fpm guidance, worked flanged-hood examples. [3] US OSHA, Fact Sheet FS-3647 "Controlling Hazardous Fume and Gases during Welding" — welding fume constituents, Cr(VI) PEL 5 µg/m³ (29 CFR 1910.1026), respiratory protection references. [4] NIOSH, "Welding Fumes and Manganese" topic page — exposure limits table (NIOSH REL 1 mg/m³ TWA / 3 STEL; OSHA PEL 5 mg/m³ ceiling; ACGIH TLV 0.02 respirable / 0.1 inhalable; IDLH 500). [5] ACGIH, Manganese documentation (2012 adoption; basis and rationale) — TLV-TWA 0.02 mg/m³ respirable (lowered from 0.2), supplementary 0.1 mg/m³ inhalable; welding-fume particle size (<4 µm, respirable); field exposures including in-helmet and confined-space measurements (0.1–1.0 mg/m³ respirable; personal exposures 0.01–4.93 mg/m³ in cited studies). [6] IARC Monographs Volume 118 (2017), "Welding, molybdenum trioxide, and indium tin oxide" and Guha et al., Lancet Oncology 2017 — welding fumes classified Group 1 (upgraded from Group 2B, 1989); sufficient evidence for lung cancer; ~11 million welders; ~110 million with welding-related exposures. [7] Keane et al., "Profiling Mild Steel Welding Processes to Reduce Fume Emissions and Costs in the Workplace" — fume generation rates (SMAW 312 mg/min, 12.6 mg/g; CMT 34.8 mg/min, 0.92 mg/g; pulsed spray 89 mg/min; globular CO₂ 387 mg/min) and comparison with EPA emission factors. [8] Keane et al., "Profiling stainless steel welding processes..." — FGR 15–230 mg/min across processes; Cr(VI) generation 50–7,800 µg/min (SMAW highest, STT lowest); Mn 50–300 µg/g. [9] Jenkins & Eagar (2005) as cited in PMC4119574 — fume particle range 0.005–20 µm; less than 10–30 % of fume mass above 1 µm; fresh-fume number median diameter 10–23 nm. [10] "Welding Fumes in a Chinese Shipyard" (PMC13030628) — ~94.2 % of fume particles <5 µm, typically 0.01–0.4 µm. [11] Ennan et al., "Particle size distribution of welding fume..." (J. Aerosol Sci. study) — three-modal distribution; dominant mode (80–90 % of particles) at geometric mean diameter ~0.22 µm. [12] US EPA shipyard welding emission factor study & AP-42 §12.19 / ARB defaults — 1–10 % of rod mass converts to fume; SMAW/FCAW default 2 %, GMAW/MIG/TIG 1 %, SAW ~0.005 %; Cr→Cr(VI) conversion 55 % (SMAW), ~5 % (GMAW). [13] "Analysis of Cyclone Pressure Drop" (Beltwide Cotton Proceedings, 2001) — 1D3D cyclone ≈ 4.5 in.wg at 3,200 fpm; 2D2D ≈ 4.4 in.wg at 3,000 fpm; velocity-head (ΔH) formulation; pressure drop independence of scale within a geometry family. [14] Iozia & Leith / Dirgo (as reviewed in Aerosol Sci. & Tech.) — cyclone cut diameter d50 models and optimization; Stairmand HE baseline (D = 0.254 m reference). [15] Senotay baghouse design guide and field notes — air-to-cloth ratio bands by industry (cement 2.5–4, metalworking 3.0–4.5, wood 4–6 ft/min; reverse-air 1.5–3); can velocity ≤3 ft/s preferred <2.5, <2.0 for fine dust; pulse pressure 80–100 psi (90 optimal); pulse interval 8–12 s / 100–150 ms; case study A/C 6.2→4.1 ft/min = +34 % bag life, −28 % Δp, ≈47k/yr maintenance saving; over-pulsing (<5 s) = up to 40 % bag wear. [16] Rough Logic dust-collector air-to-cloth reference — shaker 2–3, reverse-air 1.5–2.5, pulse-jet 3–5 (conservative default; EPA APTI lists 6–15 for some dusts), cartridge 0.5–1.5 ft/min; blanked-bag derating example (30 of 200 bags off-line = +18 % duty on the rest). [17] NFPA 660, Standard for Combustible Dusts and Particulate Solids (2025 first edition) — consolidation of NFPA 61, 484, 652, 654, 655 and 664; fundamentals in Chapters 1–10; commodity chapters thereafter (as described in NFPA Journal, May 2025). [18] NFPA 654 (2020 edition) product documentation — additive-manufacturing and ultrafine-powder provisions added; DHA deadline alignment; OSHA Combustible Dust NEP reference; 1/32-inch layer criterion as enforced through housekeeping requirements; portable vacuum/static provisions. [19] Dust explosion parameter references: Kst classes St 0–St 3 (1–200 / 201–300 / >300 bar·m/s); representative Kst/Pmax values (wood 100–200 / 8–10 bar; aluminium <75 µm 400–700+ / 11–13; magnesium 500+; titanium ~412); MIE (aluminium <1 mJ; cornstarch 40–60 mJ; coal 60–100 mJ; lycopodium 1–3 mJ reference); cubic law K_{st} = (dP/dt)_{max} V^{1/3} per ISO 6184-1 / EN 14034 / ASTM E1226 (20 l sphere and 1 m³ vessel); test concentration sweep 60–1,000 g/m³. [20] The Factories Act, 1948 (India), §13 (ventilation and temperature) and §14 (dust and fume) — quoted clause; DGFASLI/state rules as the enforcement framework. [21] Silica exposure studies in India: "Assessment of silica dust exposure profile... Indian sandstone mine workers" (Am. J. Ind. Med. 2020) and "Exposure profile of respirable crystalline silica in stone mines in India" — DGMS limit 3 mg/m³ respirable dust at ≤5 % free silica → 0.15 mg/m³ silica; OSHA PEL 0.05 mg/m³; ACGIH TLV 0.025 mg/m³; measured RCS 0.12–0.17 mg/m³; silicosis prevalence 12–30 %; recommendation to lower the Indian standard. [22] CDC/NIOSH "Recommended Industrial Ventilation Guidelines" (abrasive-blasting operations) — 500 ft/min indraft through enclosure openings; 200 cfm/ft² with curtains; duct velocity 3,500 fpm minimum; air-inlet velocities. [23] MEPMate dust-collector duct references — design transports in metric (fume ~12 m/s; wood 18; grain/paper 20; heavy metal 23) and the round-down rule. [24] HVAC Systems Encyclopedia (industrial exhaust duct design) — ACGIH minimum transport velocity table in fpm, velocity-pressure worked examples, design checklist (junction velocity rise ≤20 %, 30–45° branch entries, expansion angles, 1–2 % slope, bonding/grounding). [25] AMCA fan application literature (e.g., AMCA 201 Fan Laws; 210/ISO 5801 test standards; system-effect guidance). [26] ISO 12759 (fan efficiency classification, 2010 + parts 2019), EN 17166:2019 and EU Regulation 327/2011 — efficiency categories (static/total), fan efficiency grades (FEG), application range 125 W–500 kW. [27] US DoE Motor Systems Tip Sheet #11 and Trane Engineers Newsletter — affinity laws; 16.4 kW → 2.8 kW at half speed (82.9 % reduction) with real efficiencies; 12.5 % power at half flow; VSD system caveats. [28] Indian market listings (IndiaMART, TradeIndia, manufacturer quotes, 2026): extraction arms ₹23,000–50,000; portable extractors ₹40,000–85,000; cyclone collectors ₹35,000–2,00,000; silo-top filters ₹56,000–59,000; cartridge elements ~₹1,500; wood-dust bag units ₹1,40,000–5,40,000; baghouse units ~₹4,50,000; 5 HP fume systems ~₹85,000. [29] Industrial monitoring reference consolidations — variable-torque load exponents (power ∝ N³ ideal; field exponents 2.3–2.6 with static components); duty-cycle energy method; VFD loss allowances (2–5 % at full load). [30] Claight/market compilations (2025–2030): industrial dust collector market ≈ 9.78 bn (2025) → 12.5 bn (2030) at ~5.1 % CAGR, APAC largest; dust control systems ≈ 21.9 bn → 29.5 bn at ~6.15 %; wet dust collectors 2.0 bn → $2.71 bn at ~6.3 %.

Note on sources: exposure limits, Kst values, emission factors and the DoE/ASHRAE/ACGIH figures are engineering references with long pedigrees; where a value is a range, treat it as the range. Every worked example in this guide is arithmetic performed on its stated inputs — re-run it on your own flows, dusts, tariffs and shifts before spending money, and where a dust is combustible, the governing NFPA 660 requirements and a competent Dust Hazard Analysis supersede any general advice here. Ventilation protects people; compute like someone's lungs depend on it, because they do.

More FabFlow blog posts