Settling and Particle Removal

A particle is captured if it reaches the floor before the water reaches the weir.

Before this Lesson

A clarifier is letting too much silt slip over its weir, and there's money to make it half again as big. You could spend the concrete on depth or on width. Either way, the tank holds 50% more water, and every drop stays in it 50% longer.

Only one of the two catches more silt. To see which, follow a single grain through the tank.

A race to the floor

Start with the simplest tank that still behaves like a real one, the ideal settling basin. Water flows evenly from one end to the other at one steady speed. Particles enter spread evenly over the depth, each sinks alone at its own constant speed, and one that reaches the floor stays there.

Follow a particle that enters at the very top. The water carries it toward the outlet while it sinks, so it's a race. The water moves at the Handbook's horizontal velocity vh=Q/Axv_h = Q/A_x, the flow over the cross-section WhWh, so it crosses the length LL in LWh/QLWh/Q, the detention time θ\theta from Sizing Treatment Units. The particle sinks at its terminal settling velocity vtv_t, so it needs h/vth/v_t to reach the floor. It's caught if it gets there first:

hvt≤θ⟹vt≥hθ=h QLWh\frac{h}{v_t} \le \theta \quad\Longrightarrow\quad v_t \ge \frac{h}{\theta} = \frac{h\,Q}{LWh}

The depth cancels, and what's left is the overflow rate:

vt≥QLW=QA=vov_t \ge \frac{Q}{LW} = \frac{Q}{A} = v_o
vh h L Inlet zone Outlet zone Settling zone Sludge zone Weir vt = vo vt > vo vt < vo
Three particles entering an ideal basin at the surface
Each particle drifts forward with the water and sinks at its own speed. The one settling at exactly vov_o lands at the far corner.

Make the tank twice as deep at the same flow, and a particle at the surface has twice as far to fall. But the water, now passing through twice the cross-section, moves at half speed and gives it twice as long. The two cancel exactly. Make the tank wider or longer instead, and the particle gets more time with no farther to fall.

That settles the hook. The deeper tank catches exactly what the old one did. The wider tank has 1.5 times the area, so its overflow rate drops to two-thirds, and it now catches every particle down to two-thirds of the old cutoff speed.

The Handbook's Clarifier section calls vov_o the critical settling velocity: the terminal settling velocity of the smallest particle that is 100% removed. Anything that fast lands even from the surface, so from any height.

Slower particles, lower starts

A particle slower than vov_o misses if it starts at the top. But particles enter at every height, and a slow one that starts low enough still lands. In the time θ\theta it has, it falls vtθv_t\theta, so it's caught if it enters within hp=vtθh_p = v_t\theta of the floor.

vh h L Inlet zone Outlet zone Sludge zone Weir hp vt < vo escapes caught
Three identical slow particles entering at different heights
All three sink at the same speed, so their paths are parallel. Only the ones entering below hph_p reach the floor.

With particles spread evenly over the depth, the share that start that low is hp/hh_p/h. And since h=voθh = v_o\theta:

fraction removed=hph=vt θvo θ=vtvo\text{fraction removed} = \frac{h_p}{h} = \frac{v_t\,\theta}{v_o\,\theta} = \frac{v_t}{v_o}

A particle settling at 60% of the overflow rate is 60% removed. The ratio only works below vov_o. A particle faster than vov_o lands from every height, so it's 100% removed, and a ratio above 1 just means it had time to spare. The Handbook defines vov_o but doesn't print this fraction. It follows from the same race.

Most of the work is units. Put both speeds in the same ones first. In SI, m³/m²·d is simply m/d. In U.S. units, divide gpd/ft² by 7.481 for ft/day, and multiply ft/sec by 86,400 for ft/day.

Worked Example Removal in a rectangular basin
1 of 3

A rectangular sedimentation basin is 100 ft long, 25 ft wide, and 12 ft deep, and it treats 2.0 MGD. What share of particles settling at 0.0008 ft/sec does it remove?

Flow and area decide, not depth

Removal hangs on vt/vo=vtA/Qv_t/v_o = v_t A/Q, so you can judge most changes to a basin without a calculator. A change in the flow or the plan area shifts it in proportion, up to the 100% cap, and nothing else does: not the depth, not the detention time, and not whether the basin is long and narrow or short and wide.

Allen Hazen pointed this out in 1904, and it's why some plants stack inclined plates inside a basin. Each plate is another floor for particles to land on, which adds settling area without a bigger footprint.

Real tanks still need some depth. It holds the sludge until it's scraped out, and it keeps the water near the floor slow enough not to scour settled particles back up. And particles that stick together grow as they sink, so for them more time does help, as the last section shows.

Checkpoint
1 of 2

An ideal basin removes 70% of the particles of a certain size. The plant then sends twice the flow through it. What share of those particles does it remove now?

A.
70%, since the particles settle as fast as before
B.
35%
C.
49%
D.
It can't be found without the basin's depth.

How fast a particle falls

So far, vtv_t has been handed to you. Drop a grain of fine sand into a glass of still water, and it speeds up for an instant, then sinks at a steady speed. Three forces set that speed.

d FB FD W vt
The forces on a sinking sphere

Its weight WW pulls down, and the water pushes up with a buoyant force FBF_B, the weight of the water the grain displaces. What's left is its weight in water, (ρp−ρf) g πd3/6(\rho_p - \rho_f)\,g\,\pi d^3/6 for a sphere, with ρp\rho_p the particle's density and ρf\rho_f the water's. A particle as dense as water never settles, however big it is.

Drag, FDF_D, grows with speed until it carries the rest. For a small, slow sphere, the Handbook's drag force FD=CDρv2A/2F_D = C_D\rho v^2 A/2, with its laminar CD=24/ReC_D = 24/\mathrm{Re}, Re=vtρd/μ\mathrm{Re} = v_t\rho d/\mu, and the sphere's projected area A=πd2/4A = \pi d^2/4, works out to FD=3πμd vtF_D = 3\pi\mu d\,v_t. Set that equal to the weight in water, and you get Stokes' law. The Handbook's Settling Equations print it in two forms, the second for a particle given by its specific gravity:

vt=g(ρp−ρf)d218μ=gρf(SG−1)d218μv_t = \frac{g(\rho_p - \rho_f)d^2}{18\mu} = \frac{g\rho_f(SG - 1)d^2}{18\mu}

Weight grows as d3d^3, but this drag only as dd, so vtv_t grows as d2d^2. Halve the size, and the particle falls four times slower. A 10 μm silt grain falls 100 times slower than a 100 μm sand grain, which is why coagulation and flocculation work so hard to grow particles.

Temperature enters through μ\mu, which falls as water warms. By the water table from Fluid Properties, μ\mu at 5°C is 1.7 times its value at 25°C, so a grain sinks 1.7 times slower in winter.

Water at typical basin temperatures
Temp. (°C)ρ\rho (kg/m³)μ\mu (Pa·s)ν\nu (m²/s)
51000.00.0015180.000001518
10999.70.0013070.000001306
15999.10.0011390.000001139
20998.20.0010020.000001003
25997.00.0008900.000000893

Source: FE Reference Handbook, Properties of Water (SI Metric Units), selected rows

Use the dynamic viscosity μ\mu. The ν\nu column has no place in Stokes' law, and in SI it makes vtv_t about 1,000 times too big. Keep the units consistent: dd in meters, ρ\rho in kg/m³, μ\mu in Pa·s, and g=9.81 m/s2g = 9.81\ \text{m/s}^2 give m/s. In U.S. units, use feet, slug/ft³ (the water table's lbf-sec²/ft⁴), lbf-sec/ft², and g=32.2 ft/sec2g = 32.2\ \text{ft/sec}^2 for ft/sec. Sizes come in mm or μm (1 μm is 10⁻⁶ m), and since dd is squared, a missed conversion is squared too.

Stokes' drag holds only while the water creeps smoothly around the particle, the Handbook's laminar range of Re≤1.0\mathrm{Re} \le 1.0, so check it once you have vtv_t. Big, fast particles fail. A 0.2 mm sand grain with SG=2.65SG = 2.65 in 20°C water gets 36 mm/s from Stokes, but then Re≈7\mathrm{Re} \approx 7, and the Handbook's transitional drag coefficient gives a real speed of about 26 mm/s. For sand in 20°C water, Stokes' law holds up to about 0.1 mm.

Worked Example Will the silt be removed?
1 of 4

Silt particles 0.030 mm across, with SG=2.65SG = 2.65, enter a clarifier with an overflow rate of 40 m³/m²·d. The water is at 20°C. Find their settling velocity, and decide whether they're fully removed.

Four ways to settle

Everything so far assumed discrete particles, each sinking alone at one constant speed. Whether that holds comes down to two questions: do the particles stick to each other, and how crowded are they?

dilute Solids concentration concentrated low Tendency to flocculate high Discrete Type 1 each particle alone,at a constant speed Flocculent Type 2 particles stick,grow, and speed up Hindered(zone) Type 3 particlessettletogetheras a blanket Compression Type 4 particlesrest on eachother andare squeezed
The four settling regimes
Stickiness splits dilute suspensions in two. Crowding takes over as the solids get more concentrated.
  • Discrete (Type 1): dilute particles that don't stick, like sand and gravel. Each keeps its own speed, so Stokes' law and vt/vov_t/v_o apply as taught.
  • Flocculent (Type 2): dilute but sticky, like the organic solids in raw sewage. Particles collide, grow, and speed up as they sink, so here a deeper tank does catch more.
  • Hindered, or zone (Type 3): so crowded that the particles sink together as one blanket with a sharp top edge, slowed by the water squeezing up past them, like the mixed liquor of bacteria and water that leaves an aeration basin.
  • Compression (Type 4): so crowded that the particles rest on each other and settle only as the weight above squeezes water out, like sludge piled deep enough to thicken.

The Handbook doesn't name these regimes, so they come from standard practice.

Checkpoint
1 of 2

Which pairings of a treatment unit with the settling regime that governs it are correct?

A.
Sand and gravel in a grit chamber: discrete
B.
Raw sewage solids in a primary clarifier: compression
C.
Mixed liquor in an activated sludge clarifier: hindered (zone)
D.
Sludge in a gravity thickener: flocculent
E.
Alum floc in a water plant's sedimentation basin: flocculent

Recap

  • In an ideal basin, a particle is fully removed if vt≥vo=Q/Av_t \ge v_o = Q/A. The Handbook calls vov_o the critical settling velocity.
  • A slower particle is removed in the fraction vt/vov_t/v_o, never more than 100%. Put both speeds in the same units first.
  • Ideal removal depends on Q/AQ/A, not depth or detention time. Doubling the flow halves the removal of slow particles, and doubling the area doubles it, up to 100%.
  • Stokes' law is vt=g(ρp−ρf)d2/(18μ)v_t = g(\rho_p - \rho_f)d^2/(18\mu), with dd in meters or feet, the dynamic viscosity, and Re≤1.0\mathrm{Re} \le 1.0. It overstates the speed of large, fast particles.
  • Grit chambers are discrete settling, primary clarifiers flocculent, activated sludge clarifiers hindered (zone), and thickeners compression.