Wet Venturi Scrubber Particle Removal Simulator

A Venturi wet scrubber forces high-velocity flue gas through a narrow throat where water is injected, so dust and PM impact onto the droplets and are washed out. This simulator runs the Calvert/Yung model in real time: change particle diameter, gas velocity, L/G ratio and throat length to see how droplet diameter, Stokes number, penetration, removal efficiency and pressure drop respond — the first sizing pass for an air-pollution control device.

Parameters

Particle diameter d_p

μm

Geometric mean diameter (PM2.5 ≈ 2.5μm, PM10 ≈ 10μm)

Gas velocity v_g

m/s

Gas velocity at the throat. 80–120 m/s is typical

Liquid-to-gas ratio L/G

L/m³

Water flow ÷ gas flow. 1.0–1.5 is standard

Particle density ρ_p

kg/m³

Soot ≈ 2000, fly ash ≈ 2500, metal dust ≈ 5000

Throat length L_t

Contact length between droplets and particles. 0.2–0.5 m typical

Gas viscosity μ_g

Pa·s

Air 20°C ≈ 1.8e-5; hot flue gas 300°C ≈ 2.9e-5

Results

—

Droplet diameter d_d (μm)

—

Stokes number St

—

Single-droplet capture (%)

—

Overall penetration P_t

—

Removal efficiency η (%)

—

Pressure drop ΔP (Pa)

—

Venturi cross-section — water spray & particle capture animation

Gas flows from left through the converging → throat → diverging duct. Sprayed droplets (blue) capture particles (yellow) by inertial impaction in the throat — green = captured, yellow = bypass.

Removal efficiency η vs particle diameter d_p

Removal efficiency η vs L/G ratio

Theory & Key Formulas

$$St = \frac{\rho_p\,d_p^{2}\,v_g}{9\,\mu_g\,d_d},\quad \eta_d = \left(\frac{St}{St+0.7}\right)^{2},\quad P_t = \exp\!\left(-K\,\frac{L}{G}\,\sqrt{\eta_d}\right)$$

K = 5000 is an empirical constant; L/G is m³ water / m³ gas (slider value in L/m³ divided by 1000). St measures particle inertia, η_d is the single-droplet inertial impaction efficiency, P_t is the overall penetration (lower = better).

$$d_d = \frac{16.4\times10^{-6}}{v_g}\times 1000\;\text{[m]},\quad \Delta P = \frac{1.2\,v_g^{2}\,(L/G)\,\rho_L}{1-(L/G)}\;\text{[Pa]}$$

d_d is the Nukiyama-Tanasawa Sauter-mean droplet diameter; ΔP is the Calvert Venturi pressure-drop formula (ρ_L = 1000 kg/m³). Removal η = (1 − P_t)·100.

Wet Venturi Scrubber Particle Removal

🙋

A Venturi scrubber is the box that stops black smoke from coming out of the stack, right? What actually happens inside?

🎓

Exactly — it's the workhorse in flue-gas cleanup for coal-fired power, steel mills and waste incinerators. The mechanism is surprisingly mundane: the flue gas is pushed through a narrow "throat" section and accelerated to 80–120 m/s. Water is sprayed in from above, and from the gas's point of view it suddenly hits a wall of slow, huge droplets. The particles have mass, can't follow the streamlines, and impact onto the droplets by inertia. That is inertial impaction — the headline mechanism in the Calvert/Yung model.

🙋

"Hit them with droplets" — got it. So when I drop the particle diameter from 5μm to 0.5μm and efficiency collapses, is it because the particles are too light to leave the streamlines?

🎓

Exactly. Stokes number goes as d_p², so dropping from 5μm to 0.5μm cuts inertia by 100×. Single-droplet capture η_d = (St/(St+0.7))² collapses, and the overall penetration P_t = exp(−K·L/G·√η_d) flattens against its ceiling. You'll see the cliff at the left end of Chart 1. That is why the 0.1–1μm band is called the Greenfield gap: Venturis alone struggle there, so engineers add an upstream electrostatic precipitator, dose agglomeration aids, or add a separate absorption tower for water-soluble gases.

🙋

Can't we just keep raising L/G until everything is captured? Chart 2 seems to plateau around 2 L/m³.

🎓

Three walls stop you. First, water-pump power scales linearly: triple L/G, triple pump kW. Second, channelling — above L/G ≈ 3 the water no longer spreads uniformly, so the extra flow does less work than the model promises. Third, slurry. Recycled liquor concentration rises, blowdown grows, and effluent treatment cost catches up with the gain. Practical designs sit at L/G = 1.0–1.5 and only push to 2.0–2.5 for hard PM2.5 service.

🙋

What about the pressure drop? The default ΔP is about 11500 Pa = 11.5 kPa. Is that high?

🎓

Yes — the default ΔP of 11.5 kPa is medium-to-high for a Venturi. PM10 service runs comfortably at 2–5 kPa, but reaching 99% on PM2.5 typically wants 8–15 kPa. ΔP is the dominant OPEX driver because it goes straight into the blower. So the proper design path is "target efficiency → required ΔP → throat area (= gas velocity)" — pick the geometry first, then tune L/G and droplet size. Calvert himself published η = 1 − exp(−c·ΔP^n) — empirically, doubling ΔP improves the 0.5μm efficiency by roughly one order of magnitude.

🙋

One more — what is the K = 5000 constant inside this simulator? Some textbooks quote K = 200.

🎓

K is an empirical lumped constant. It depends on which units you use for L/G (L/m³ vs m³/m³), the droplet-size distribution and the throat geometry, so it spans 1000 to 10000 in the literature. This tool converts L/G into m³/m³ (÷ 1000) and then picks K = 5000, the central value for circular Venturis with a Nukiyama–Tanasawa droplet. The K = 200 you find in textbooks is the version that keeps L/G in L/m³ directly. For a real machine, calibrate K against measured data from a similar unit and treat this simulator as an early-design sensitivity tool.

Frequently Asked Questions

Because the Stokes number St that drives inertial impaction scales with d_p², the Calvert/Yung model predicts a steep efficiency drop for 0.1–0.5μm particles — the so-called Greenfield gap. Around 0.3μm, single-mechanism efficiency falls to 20–60%. To still reach 99% overall removal you must push L/G to 2–3 L/m³, gas velocity above 100 m/s and the throat length to 0.3–0.5 m. Below 0.1μm Calvert alone is not enough; you typically add an upstream ESP, agglomeration aids or a downstream baghouse.

In theory P_t = exp(−K·L/G·√η_d) drops exponentially as L/G rises, but in practice three limits appear. (1) Pump power and water-treatment cost grow linearly. (2) Above L/G ≈ 3 L/m³ the flow channels and capture no longer follows the model. (3) Slurry concentration rises, increasing blowdown and effluent treatment load. Industrial designs typically sit at L/G = 1.0–1.5 L/m³ and only push to 2.0–2.5 for demanding fine-particle service.

Calvert's correlation is η = 1 − exp(−c·ΔP^n) with c, n depending on L/G and particle size. Empirically, doubling ΔP raises the 0.5μm capture efficiency by roughly one order of magnitude (90 → 99%). But ΔP drives blower power, and going from 5 kPa to 10 kPa roughly doubles fan brake power. The design order is therefore: target efficiency → required ΔP → gas velocity (throat area). PM10 service runs at 2–5 kPa, PM2.5 service at 6–12 kPa.

K is an empirical constant that ranges from 1000 to 10000 depending on the L/G unit (m³/m³ vs L/m³), the droplet-size distribution and the throat geometry. This tool converts L/G to m³/m³ (= L/m³ ÷ 1000) and uses K = 5000, a central literature value for circular Venturis with a Nukiyama–Tanasawa droplet size. For a real machine, calibrate K against measured data from a similar scrubber. Treat this simulator as an early-design sensitivity tool, not a procurement-grade model.

Real-World Applications

Coal-fired and biomass power plants — fly-ash removal: The world-standard layout is "ESP + WFGD (wet flue-gas desulfurization)". An electrostatic precipitator drops the coarse fraction (> 10μm), and a wet scrubber polishes the remaining PM2.5 while absorbing SO₂. The scrubber stage runs L/G = 1.0–1.5 and ΔP = 3–5 kPa, capturing fine particles and absorbing SO₂ with a limestone slurry simultaneously. Use this tool with PM2.5 (d_p = 2.5μm) and see how η jumps 1–2 orders of magnitude as L/G goes from 1.0 to 1.5.

Steel mills — converter and blast-furnace gas cleaning: Converter gas is hot and loaded (10–30 g/m³ of dust), so high-energy Venturi scrubbers (ΔP = 15–25 kPa, v_g = 120–150 m/s) are required to deliver > 99.9% removal. The capture water is iron-rich, so it is thickened, dewatered and recycled to the sinter plant. Push v_g in this tool and watch ΔP grow as v_g².

Waste incinerators — HCl and heavy-metal control: Municipal and industrial waste incinerators must remove HCl, mercury and dust simultaneously, so wet + dry trains are common. The Venturi handles water-soluble gases and coarse particles, with a downstream baghouse and SCR for fine PM and NOx. Combined with a quench stage, the Venturi rushes the gas through the dioxin re-formation window (200–400°C) before any reformation can occur.

Chemical and semiconductor exhaust treatment: Process drains from catalytic reactors, acid gases (HCl, HF, NH₃) and CVD/etch exhausts often need wet scrubbing as a first choice. Semiconductor fabs treat PM + gas + VOC with multi-stage scrubbers plus activated carbon, and route the rinse water back into ultrapure-water makeup in a closed loop. This tool covers only the particle side, but it is a useful starting point for sizing the scrubbing stage of such combined trains.

Common Misconceptions and Pitfalls

The biggest pitfall is copying the Calvert constant K straight from a textbook without checking units. The K = 5000 used in this tool assumes L/G has been converted to m³/m³. Other texts publish K = 200 because they keep L/G in L/m³. Mix the two and you are three orders of magnitude off. Always pair "the definition of K" with "the unit of L/G", and for a real machine calibrate K by back-fitting an existing scrubber's measured data.

Second, describing efficiency by Stokes number alone. Calvert only models inertial impaction. Below ~0.5μm, Brownian diffusion, interception, electrostatic effects and thermophoresis become non-negligible. In particular, below 0.1μm the Stokes number is tiny but diffusion efficiency rises sharply, and the combined efficiency forms the familiar U-shape vs particle diameter. This tool is intended for d_p ≥ 0.5μm; in the 0.1μm range it will under-predict the real capture.

Third, using the computed ΔP directly for fan selection. The ΔP here is close to a dry-throat value. The real machine adds losses from the mist eliminator, the converging/diverging sections, water injection head and recirculation pump head — totalling 1.3–1.7× the bare value. Scale build-up and droplet carry-over also degrade ΔP with time, so allow at least 30% margin when sizing the induced-draft fan.

Wet Venturi Scrubber Particle Removal Simulator