A mist eliminator removes liquid droplets, not gas-phase pollutants. Ask what is a mist eliminator and the working answer is a liquid-gas separator: droplets leave with the gas, the element takes them back out, and dissolved gas stays with the scrubber chemistry upstream. A plant moving 3,000 m³/h of gas with entrained droplets is the case that makes the rest concrete.
Key Takeaways
- A mist eliminator removes liquid droplets, not gas-phase pollutants. The droplets were already liquid before the gas reached the element.
- Four capture mechanisms divide by droplet size. Cut-off diameter moves as gas velocity moves, so no single size rating travels.
- 3,000 m³/h turns into face velocity, then into cross-section and diameter. Five face velocities give five different tower sizes.
- A fixed 1.2 m tower runs 0.74 m/s, below every type’s lower bound. Column size can rule out a whole family before efficiency is discussed.
- Measured pressure drop climbs near 1.51 power, not the 2.00 a square law implies. Fan power follows that difference.
- Sizing, materials, and guarantees belong to a separate decision. The type choice ends where the calculation begins.
The Droplet Carryover a Mist Eliminator Removes
Carryover reaches the top of a vessel because something upstream lifted liquid into the gas. A boiling or sparged layer throws droplets into the vapor space, and the rising gas carries them faster than gravity returns them.
How Entrainment Puts Liquid Where It Does Not Belong
Entrainment is mechanical. Gas shears droplets off a liquid surface, a bubble bursting there throws a finer spray, and the upward flow carries both. The mass that matters is the fine fraction. Coarse droplets fall back within a short distance, while droplets of a few microns stay suspended the full vessel height. Outlet loading tracks that tail, so a separator sized on the visible splash misses the part that leaves.
Size bands describe where a separator earns its place. A separator carrying droplets of 250–380 µm out of a bubble column faces a different job from one meeting an acid mist of a few tenths of a micron, and the two rarely share an element.
Product Recovery and Emission Control as Two Separate Jobs
The same element serves two commercial purposes, and the target moves with them. Recovering an expensive solvent rewards high capture because each captured kilogram returns to the process. Controlling an emission rewards a plume that has gone, and the acceptable outlet loading comes from a permit instead of from process economics. Placement follows the motive. Fiber beds sit downstream of other control devices in chemical, plastics, asphalt and surface coating plants, where the job is eliminating a visible plume. Where recovery is the motive, captured liquid has to drain back into the process, which makes drainage geometry a recovery decision.
Demister as the Older Name for the Same Device
Demister and mist eliminator name the same equipment. A demister pad is a mesh pad, and fiber-bed scrubbers list mist eliminators among their own other names. Nothing in a specification changes because the older word is used. Whether the problem is entrained liquid at all is the first thing to settle. A plume that persists, a drain leg that runs continuously, or a fan with a water trap all point to carryover; a reading that tracks gas-phase concentration points upstream to the scrubber chemistry instead.
The Four Ways Droplets Get Caught
Four mechanisms remove droplets from a gas stream, and droplet size decides which one carries the duty. A mechanism only works when the droplet’s own inertia, width or random motion brings it into contact with a surface.
Inertial Impaction Takes the Droplets Above a Few Microns
Gas flowing through a mesh or a vane must change direction around every filament and every bend, and a droplet heavy enough cannot follow that turn. It leaves the flow line and strikes the surface. Impaction carries most of the separation for wire mesh. Interception and diffusion add little to a knitted pad’s efficiency, which is why mesh and vane both count as inertial collectors and why both are rated on velocity.
Interception and Diffusion Reach the Smallest Droplets
Interception catches a droplet that follows the gas path yet is wide enough to graze the collecting surface. The droplet’s centre misses the obstacle while its edge does not. Diffusion works below that. Droplets of a fraction of a micron are small enough for gas molecules to knock them off course, so they wander across flow paths and land on a filament through random motion. A packed fiber bed removes pollutants through inertial and diffusional interception together, and the diffusion half is what reaches the sizes a mesh pad passes.
The wire does the work at a scale worth picturing. Knitted pads are built from wires of 0.14–0.28 mm, packed to a surface area of 236–868 m²/m³ inside a void fraction of 97–98.3%. A droplet crossing that lattice meets thousands of wire surfaces per metre of travel.
Gravity Settling Rarely Decides on Its Own
Gravity works before the gas reaches the element, not inside it. In a bubble column, part of the large droplets returned to the pool under gravity while most continued upward toward the separator, so the element never sees the coarsest fraction. Inside the element, gas velocity can cancel gravity outright. Above a critical velocity the free drainage of the separator is impeded by the rising gas, and at sufficiently high velocity the effect of gravity and surface tension is overcome by the pressure of the gas. Settling stays a vessel-space margin; the fabric of the element does the separating.
Cut-Off Diameter Moves When Gas Velocity Moves
A cut-off diameter marks the fine end of a separator’s range, and it moves with gas velocity rather than sitting as a fixed property. Two packs rated the same at one speed separate cleanly at another, which is why the rating belongs to a velocity as much as to a device.
The numbers for a wave-plate pack run 9.4 µm at 2 m/s, 6.4 µm at 4 m/s and 4.1 µm at 8 m/s. Faster gas drives droplets harder toward the wall and strips captured liquid back off, and the second effect is why an optimum velocity range exists at all. Interpolated to 3 m/s, the cut-off lands near 7.5 µm, so a duty specified at one velocity tells you nothing about the same element at another.
Packing density shifts the same curve. Raising surface area, packing density or superficial gas velocity pushes removal efficiency up; raising void fraction or wire diameter pulls it down. A pad run at 0.109 m/s and the same pad run at 0.118 m/s are not the same separator on paper, even though both sit inside one nominal band. A rating quoted without its velocity describes one operating point and travels no further. Read the cut-off, the velocity and the element together, then check whether the duty sits where that pair applies.
What 3,000 Cubic Meters per Hour Demands
One flow rate and one face velocity give a cross-section, and a cross-section gives a diameter. Run the same 3,000 m³/h through the velocity band each type needs and the tower size moves by half a metre.
One Duty, Five Face Velocities
Mesh pads run at bed velocities of 2–4 m/s, and vane packs run at 2.5–5.0 m/s. Those four numbers bracket the whole decision, because a velocity outside its own band pushes a given type past what its geometry supports. Take the bounds as design points. Mesh at 2.0 m/s, 3.0 m/s and 4.0 m/s; vane at 2.5 m/s and 5.0 m/s. Every one of those five sits inside its own band, so none of them is a stretch of the published range.
Face Velocity Turns Into Cross-Section and Diameter
Divide the flow by the velocity and the cross-section falls out. At 3,000 m³/h, mesh at 2.0 m/s needs 0.417 m², which is a 0.73 m circle. Raising mesh velocity to 3.0 m/s cuts the area to 0.278 m² and the diameter to 0.59 m; at 4.0 m/s it is 0.208 m² and 0.52 m.
Vane behaves the same way against a higher band. At 2.5 m/s the 3,000 m³/h duty needs 0.333 m², a 0.65 m circle. At 5.0 m/s it needs 0.167 m² and 0.46 m across. The spread from largest to smallest is 0.73 m against 0.46 m. A factor of 1.6 in diameter is a factor of 2.5 in cross-section, and that is the whole cost of running a higher velocity through the same duty.
| Element | Face velocity | Cross-section | Equivalent diameter |
|---|---|---|---|
| Mesh | 2.0 m/s | 0.417 m² | 0.73 m |
| Mesh | 3.0 m/s | 0.278 m² | 0.59 m |
| Mesh | 4.0 m/s | 0.208 m² | 0.52 m |
| Vane | 2.5 m/s | 0.333 m² | 0.65 m |
| Vane | 5.0 m/s | 0.167 m² | 0.46 m |
A Fixed 1.2-Meter Tower Runs Below Every Lower Bound
Reverse the question. Suppose the tower already exists at 1.2 m diameter. Its cross-section is 1.131 m², and 3,000 m³/h through that area gives 0.74 m/s. That single number settles nothing about efficiency and everything about feasibility. Mesh needs at least 2.0 m/s and vane needs at least 2.5 m/s, so the existing tower delivers roughly a third of the velocity mesh wants and under a third of what vane wants.
A separator dropped into that tower runs below its own band. Droplets drift through the element too slowly for impaction to do the work, and the fix is a smaller flow area rather than a different element. That means a reducing section, a smaller vessel, or a decision that this duty belongs to a larger flow than the tower can serve at any useful velocity.
| Fixed geometry | Value |
|---|---|
| Tower diameter | 1.2 m |
| Cross-section | 1.131 m² |
| Face velocity at 3,000 m³/h | 0.74 m/s |
| Mesh lower bound | 2.0 m/s |
| Vane lower bound | 2.5 m/s |
The 0.74 m/s result is the strongest single fact in this decision. It says column geometry can rule out an entire family before anyone discusses efficiency, pressure drop or price, and it is why the duty and the vessel have to be read together.
Mesh, Vane, and Fiber Bed Against the Same Duty
Four rows in a selection table place the three types against one duty, and the 3,000 m³/h case says which row each one occupies. Source: Al-Dughaither et al., 2010, Table 1, citing Ziebold 2000.
Four Rows in One Selection Table
| Style | Collecting fiber diameter (µm) | Bed velocity (m/s) | Pressure drop (mm H₂O) | Particle size collected (µm) |
|---|---|---|---|---|
| Brownian fiber beds | 8–10 | 0.05–0.25 | 100–450 | <0.1–3 |
| Impaction fiber beds | 10–40 | 1.25–2.5 | 100–250 | 1–3 |
| Mesh pads | 100–300 | 2–4 | 10–75 | 2–20 |
| Vane separator | >300 | 2.5–5.0 | 3–25 | >20 |
*Equipment selection versus mist particle size (Ziebold, 2000).*
Two rows carry fiber beds and the split between them matters. Brownian beds take the finest droplets at 0.05–0.25 m/s and pay 100–450 mm H₂O for it, while impaction beds run 1.25–2.5 m/s at 100–250 mm H₂O and stop at 1–3 µm. One label covers both ends, and the choice inside the family is its own decision.
Knitted Mesh Numbers From the Bench
A knitted pad is a bed of fine wires interlocked by knitting, commonly built 10–15 cm deep on a 15 cm test column, in 316L stainless steel. Depth sets how many chances a droplet gets at a wire, and the depth range is narrow because the pressure cost rises with every added centimetre.
The working range for a mesh pad runs 236–868 m²/m³ of surface area inside a void fraction of 97–98.3%, with wire diameter 0.14–0.28 mm and packing density 130–240 kg/m³. Pads at the top of that surface-area band catch more and cost more pressure drop on the same gas. Against the 3,000 m³/h duty, a mesh row needs 0.52–0.73 m of diameter depending on where in the 2–4 m/s band it runs.
The Vane Geometry Behind the Performance
A wave-plate pack is described by four dimensions, and each one moves capture or drainage. Plate spacing L sits at 110 mm, the channel angle θ at 100°, the hooked vane length l at 5 mm and pack height H at 80 mm.
Those four numbers set what the pack can catch. Wider spacing lets a larger droplet through the turn; a longer hooked vane and a narrower angle give the gas more turn per unit depth. H at 80 mm bounds how many bends fit, and each added bend buys capture against pressure drop. Against the same 3,000 m³/h, the vane row needs 0.46–0.65 m of diameter across the 2.5–5.0 m/s band.
Where Each Type Rules Itself Out
Each type excludes itself under a condition that can be checked before any calculation. Those conditions are worth memorising, because each one turns a comparison into a shortlist without a single number being computed.
Mesh rules itself out on a dirty gas. A pad whose void fraction sits above 97% has little room for deposits, so insoluble material closes the path the gas was using. Vane rules itself out at low velocity: below 2.5 m/s the gas cannot throw droplets onto the plates, and the pack passes them. Fiber bed rules itself out on solids and temperature, because insoluble and coarse PM will clog the fiber bed with time, and plastic fiber beds are generally restricted to operate below 60°C (140°F).
Against the 3,000 m³/h duty, the 1.2 m tower fails all three at 0.74 m/s. A vessel sized from the table instead lands at 0.46–0.73 m depending on the type, which is why the four columns have to be read as one set.
The Pressure Drop and Efficiency Numbers That Come With Conditions
Measured pressure drop climbs more slowly than a square law predicts, and every efficiency figure belongs to the test it came from. Both facts change what a fan has to deliver and what a guarantee can promise.
Pressure Drop Grows Faster Than Velocity
A wave-plate pack measured at 2, 4 and 8 m/s shows 108 Pa, 276 Pa and 882 Pa. Fit those three points and the exponent lands near 1.51, well below the 2.00 a square law implies. That gap is worth money. Doubling velocity multiplies pressure drop by about 2.85 rather than 4.0, so a fan sized on the square law comes out oversized at the top of the range. The exponent also flags a second effect: at the highest velocity the measured rise steepens, because captured liquid starts coming back off and the pack is no longer doing clean-gas work.
Fan power follows pressure drop times flow. For the 3,000 m³/h duty the absolute pressure drop depends on where in the band the pack runs, and the fitted exponent is what lets you move from one velocity to another without measuring again.
Efficiency Percentages That Only Hold in Their Own Test
Efficiency numbers describe a unit against its own conditions, and a set measured on one pack does not travel to another. A figure quoted without its velocity and liquid load is a number without a subject. A wave-plate pack tested across water flow rates and velocities produced 50.4% and 60.8% overall collection in one condition, and 75.6% and 85.3% in another. The spread between those pairs is larger than the difference between the two models, which is the clearest possible warning that a bare percentage carries no information on its own. Water load moves the number too. Raising water flow from 0.2 to 0.8 L/min cut overall efficiency by roughly 4.2% at 4 m/s and by roughly 12.7% at 8 m/s on the same pack, and the loss grew with velocity.
Fiber beds quote a wider band for the same reason. Collection efficiencies for PM and VOC mists generally run from 70% to greater than 99%, depending on the aerosol size and the design of both the scrubber and the beds.
What Those Numbers Mean for the 3,000 m³/h Duty
Put the two together and the 3,000 m³/h case gets a working expectation rather than a promise. At the low end of the mesh band, 2.0 m/s, the pack sits where the pressure drop curve is flattest, so a 0.73 m element carries the duty at modest fan cost. At the 4.0 m/s end the same 0.52 m element saves vessel size and spends it on fan power.
The fitted 1.51 exponent means that trade is milder than the square law suggests, and the efficiency spread means it is not free. A pack near the top of its velocity band returns a share of what it caught; a pack near the bottom passes what it never touched. The band, not a single velocity, is what a specification has to name.
How These Devices Fail at the Edges of Their Range
Three failure modes mark the edge of a separator’s range, and each one has a number attached. Re-entrainment arrives at a velocity, overloading arrives at a liquid load, and plugging arrives on a schedule set by the solids.
Re-Entrainment Arrives Around 10 Meters per Second
Re-entrainment means the element caught the droplets and then handed them back. Faster gas drives droplets harder onto the collecting surface and raises capture, and past a point that same velocity tears the film off and carries it out. The threshold belongs to the geometry rather than the family. On a wave-plate pack, re-entrainment clearly occurred at a speed of 10 m/s or higher, and testing stopped short of that edge for the reason it names.
Liquid load moves the threshold down. Raising water flow from 0.2 to 0.8 L/min cost roughly 12.7% of overall efficiency at the top of the tested range against 4.2% at half that speed, so gas velocity and liquid load compound instead of adding. A thicker film explains it. More water or more velocity thickens the film on the plate, and the film is what the gas tears away.
Liquid Load Turns Grams Into Kilograms
Liquid loading arrives in grams per cubic metre and has to be turned into kilograms per hour before it means anything to a drain. The conversion is the flow rate, nothing more, and it turns a concentration into a daily handling problem. Inlet loadings run 0.2–11 g/sm³. Against the 3,000 m³/h duty that is 0.60 kg/h at the low end, 3.00 kg/h at 1 g/sm³ and 33.00 kg/h at the high end.
Those three numbers size the drain and the water supply, and they explain why the low end of a loading range is a different installation from the high end. At 33.00 kg/h the drain leg and the liquid distribution matter as much as the element.
Plugging Follows the Wrong Selection
Plugging is the slow failure, and it starts with a choice that ignored the solids. Insoluble and coarse PM will clog the fiber bed with time, which makes a fiber bed a poor first choice for a dusty gas regardless of how fine its droplets are. A mesh pad closes through a different route with the same result. A pad whose void fraction sits above 97% has little tolerance for deposits, so material that will not dissolve occupies the void space and shuts the path the gas was using. Pressure drop is the early signal. Fiber-bed rows already sit an order of magnitude above vane rows on pressure drop, so a bed that is partly closed shows up as fan power climbing while the outlet stays clean.
Read against the 3,000 m³/h duty, the edges line up as numbers. A 0.46 m vane element runs closest to the 10 m/s ceiling, a 0.73 m mesh element sits furthest from it, and either one plugs sooner on a gas carrying solids at 33.00 kg/h of liquid load.
Where the Decision Stops and Sizing Begins
The decision ends when the type and the velocity band are fixed. Everything after that is a calculation with its own inputs, and the 3,000 m³/h case shows exactly where the line falls.
The Decision Path in One Pass
Run the whole chain once. 3,000 m³/h is 0.83 sm³/s, which sits inside the 0.5–47 sm³/s range fiber beds cover. Face velocity turns that flow into 0.417 m² and a 0.73 m circle for mesh at 2.0 m/s, or 0.167 m² and 0.46 m for vane at 5.0 m/s.
Fix the tower at 1.2 m instead and the same flow gives 1.131 m² and 0.74 m/s, which is below the 2.0 m/s mesh floor and the 2.5 m/s vane floor. At that velocity no inertial element does its job, and the answer is a smaller flow area rather than a different pad.
The chain closes on three numbers: 3,000 m³/h in, a diameter between 0.46 m and 0.73 m out, and a 0.74 m/s warning for a vessel that is already built. Those three are what a supplier needs before the conversation can move to price.
Sizing, Materials, and Guarantees Belong Elsewhere
Sizing takes over from here. Vessel diameter, element thickness, support structure and the resulting pressure drop follow from the duty once velocity is fixed, and that work belongs to a sizing guide rather than to a type comparison. Material follows the chemistry of the gas and liquid over the service interval, and temperature alone removes some options before corrosion is considered. An efficiency guarantee is a commitment against one specification for one unit, and it comes from whoever will answer for the measurement.
The Inputs a Supplier Needs
An enquiry moves quickly when the inputs arrive complete. A supplier who has to ask for the flow rate has already lost a day, and one who has to ask for the droplet size has lost the basis for the choice. Six items carry the decision: the droplet size you must capture, the flow at operating temperature and pressure, the liquid load in g/m³, the allowable pressure drop, the temperature range, and whether the goal is recovery or an emission limit. With those six, a supplier answering what is a mist eliminator for your duty can place it in a row of the table and quote a diameter.
If an input is missing, record it as open and name the owner. A flow nobody measured and a droplet size nobody sampled are different problems, and only one of them is closed by a phone call. That is the point where this decision hands over, with 3,000 m³/h and a 1.2 m tower behind it and a sizing calculation ahead.
Questions the Numbers Raise
Can I keep my existing 1.2 m tower and just fit a different element?
Changing the element does not change the face velocity, because 3,000 m³/h through 1.131 m² is 0.74 m/s no matter what sits inside the vessel. The area and the flow set that number, and the element only decides what happens at the velocity it is handed. Mesh wants at least 2.0 m/s and vane wants at least 2.5 m/s, so the existing tower delivers under a third of the slower floor and under a third of the faster one.
Three routes exist and each one costs something different. Reducing the flow area with a reducing section takes the 1.131 m² down to 0.417 m², which is a 0.73 m equivalent diameter and just reaches the mesh floor at 2.0 m/s. Replacing the vessel with a larger one costs a vessel, and accepting the tower as it stands means it cannot serve this duty at any useful velocity. The first route is the cheapest of the three and the only one that keeps the existing shell.
Do I need the same face velocity for mesh and for vane?
No, because the two are not measuring the same thing. Mesh separates on wires of 0.14–0.28 mm and relies on droplets failing to follow the gas around a filament, while a vane pack separates across open channels measuring several hundred micrometres and relies on the gas turning sharply enough that inertia carries the droplet onto a plate. Different geometry produces a different velocity band, and neither band is a property of the other.
The same 3,000 m³/h duty shows the split in diameters. Mesh across 2–4 m/s needs 0.52–0.73 m, and vane across 2.5–5.0 m/s needs 0.46–0.65 m, so one duty carries two separate size ranges depending on which route is chosen. Pick a route and then take that route’s band, because mixing a mesh velocity with a vane diameter describes an element that does not exist. The tighter vane diameter also comes with a pressure drop and a re-entrainment margin that the mesh figure does not carry.
