An introduction to sedimentation theory in wastewater

treatment
Bengt Carlsson
Systems and Control Group
Uppsala University
Nov 96, rev Okt 98

Abstract
This material is made for the course \Wastewater treatment" in the Aquatic and
Environmental Engineering program. The sections which are marked with a  are
not central in the course.

Contents
1 Introduction 1
2 Discrete particle settling 3
2.1 Stokes law* . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 The surface loading rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3 Hindered settling 5
3.1 The solid ux theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.2 Veslinds formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.3 Numerical solution of the settler* . . . . . . . . . . . . . . . . . . . . . . . 7

1 Introduction
Sedimentation (settling) is the separation of suspended particles that are heavier than
water. The sedimentation of particles are based on the gravity force from the dierences
in density between particles and the uid. Sedimentation is widely used in wastewater
treatment systems. A successful sedimentation is crucial for the overall eciency of the
plant. Common examples include the removal of;
 Grit and particulate matter in the primary settling basin (settling tanks that
receive raw wastewater prior to biological treatment are called primary tanks,
forsedimentering).
 Sludge from the bioreactor (activated sludge process).
 Chemical ocs in the chemical step.

Often, the settler connected to the activated sludge process is the main bottle neck in the
plant. The seemingly simple process has proven to be the weak link in many wastewater
treatment plants.

1
The implementation of nitrogen removal in many Swedish plants emphasis the impor-
tance of the settler. The slow growth of nitrifying bacteria means that a high sludge age
is necessary in the activated sludge process. For a give volume of the aeration basin, the
sludge age may be increased by using a higher sludge concentration in the basin. How-
ever, by increasing the sludge concentration in the aeration basin, the capacity of the
settler may be reached, the sludge blanket level will then increase which nally results
in an uncontrolled sludge escape in the euent water. Hence, there is a possible con ict
between operation for good nutrient removal (high sludge age) and operation for good
sludge sedimentation. Further, nitrogen removal in the activated sludge process gives
also a risk for sludge rise in the secondary settler due to denitrication in the bottom
of the settler. The sludge may rise due to otation of solids when nitrogen gas is released.
Note also that the settler has two functions; clarication and sludge separation. That is
to remove essentially all of the solids from suspension and to concentrate theses solids
(eg for recycling to the aeration basin).
Depending on the particles concentration and the interaction between particles, four
types of settling can occur, see also Figure 1:
 Discrete particle settling. The particles settle without interaction and occurs under
low solids concentration. A typical occurrence of this type of settling is the removal
of sand particles.
 Flocculent settling. This is dened as a condition where particles initially settle
independently, but occulate in the depth of the clarication unit. The velocity
of settling particles are usually increasing as the particles aggregates. The mecha-
nisms of occulent settling are not well understood.
 Hindered settling. Inter-particle forces are sucient to hinder the settling of neigh-
boring particles. The particles tend to remain in a xed positions with respect to
each others. This type of settling is typical in the settler for the activated sludge
process (secondary clarier).
 Compression settling. This occurs when the particle concentration is so high that
so that particles at one level are mechanically in uenced by particles on lower
levels. The settling velocity then drastically reduces.

Discrete particle settling

Flocculent settling

Hindered (or zone) settling

Compression

Figure 1: Settling phenomena in a clarier

In the following, we will restrict the discussion to discrete particle settling and hindered
settling.

2
2 Discrete particle settling
Consider the settling of a discrete particle, see Figure 2. The sedimentation is obtained
by the Newton and Stokes law.
Fl
Ff Ff

Fg

Figure 2: Forces on a discrete particle, Fg is the gravitational force, Fl is the "lifting"

force from the liquid, and Ff is the frictional force between the particle and the liquid.
Newton second law gives
m dv
dt = Fg , Fl , Ff (1)
where v is the velocity of the particle, and m is the mass.
The gravity force Fg is given by
Fg = mg = pVpg (2)
where p is the density of the particle and Vp is the volume.
The lifting force Fl is given by
Fl = f Vp g (3)
where f is the density of the uid.
The frictional drag force Ff depends on the particle velocity, uid density, projected
area, and a drag coecient. The following empirical expression is used
Ff = CD Ap2 f v
2
(4)
where CD is the drag coecient1 and Ap is the projected area of the particle perpendic-
ular to the velocity.
Inserting (2), (3) and (4) in (1) yields
m dv CD Ap f v 2

dt = g(p , f )Vp , 2 (5)

In steady state ( dv
dt = 0), we have
s
v = 2g(Cp A, f )Vp (6)
D p f
1
Newtons motstandskoecient.

3
2.1 Stokes law*
For a spherical particle with diameter d, we have the volume Vp = d6 3 and a projected
area Ap = d4 2 Inserting this in (6) gives
s
v = 4g(3Cp ,f )d (7)
D f
For laminar ows it holds that
CD = R24 (8)
N
where RN is the Reynolds number.
RN = vd

f (9)
In (8), the viscosity  is introduced. This gives a measure of a uids resistance to tan-
gential or shear stress.
Inserting (8) in (7) using (9) gives Stokes law
, f )d
v = g(p 18
2
(10)

2.2 The surface loading rate
The settling velocity of a particle can be used in the design of settling (sedimentation)
basins. The key idea is to nd a lower limit on the settling velocity for the particle to
settle before it reach the outlet. Consider an ideal settling basins according to Figure 3.

Figure 3: Settling in an ideal settling basin. The basin depth is h, the surface area is A.
The time a unit element is residencing in the (ideal) settling basin is given by
T = VQ = Ah
Q (11)
where V is the basin volume, A is the surface of the basin, h is the basin depth, and
Q is the ow rate. The minimum settling velocity for a particle (entering the basin at
height h) to settle is thus given by
vmin = h=T = Q
A (12)
The ratio QA is the surface loading rate2 and is one of the key parameters in the operation
and design of settling basins. Note, however, that for normal settling basins the above
2
Ytbelastningen.

4
relation only gives a crude approximation. Typically, only about 60% of the theoreti-
cal settling capacity (12) is achieved in practice. Dierent empirical relations exist to
compensate for non ideal situations. Note that the basin depth h does not in uence
the (theoretical) minimum velocity. Common experience suggests that the basin depth
should exceed 2-3m.

3 Hindered settling
3.1 The solid ux theory
Most models for hindered settling are based on the solid ux theory. Pioneering work,
using solid ux concept for settler calculations, was done by Cloe and Clevenger (1916)
and Kynch (1952).
In general, the total ux (mass/(area time) of solids is obtained by
J = Xv (13)
where X is the solid (sludge) concentration3 and v is the settling velocity which in gen-
eral depends on X .
Now, consider a settling basin according to Figure 4.
Qin Qe

Qu

Figure 4: A settling basin. The basin has a cross sectional area A. At the bottom of the
basin, sludge is withdrawal at the rate Qu .
The total ux of solids through a segment is
Jt = Jg + Ju (14)
where Jt is the total ux, Jg is ux due to gravitational settling, and Ju is the ux
resulting from the sludge withdrawal at the bottom of the basin (often called bulk ux).
From Figure 4 we have that
Ju = vuX = QAu X (15)
3
The concentration of suspended solids is normally given as MLSS, mixed liquor suspended solids

5
where vu is the velocity resulting from the removal of sludge at the bottom of the tank..
The solid ux theory states that
Jg = vg (X )X (16)
That is, the gravitational settling velocity only depends on the local concentration of
solids. Hence, we can write the total ux as
Jt = (vg (X ) + QAu )X (17)

3.2 Veslinds formula

Several empirical relations have been suggested to describe the relation between the vg
and X . A commonly used relation is the Veslind formula:
vg (X ) = vo exp,nX (18)
where vo is the maximum settling velocity and n gives a measure on how fast the settling
velocity decreases with increasing concentration of particles. In practice, these param-
eters can be found by multiple batch settling experiments where log vg is measured for
dierent sludge concentrations. Then, vo and n can be found by a simple least squares
t to the data (linear regression). Also, several empirical relations exist where one tries
to relate parameters like SV I (sludge volume index) to vo and n.
Inserting the Veslind formula (18) in (17) yields
Jt = (vo exp,nX + QAu )X (19)
An illustration of the relation (19) is given in Figure 5.
180

160

140
J_lim

120

100
Flux

80

60

40

20

0
0 2 4 6 8 10 12 14 16
Concentration

Figure 5: Total ux as a function of the concentration.

Notice that the ux curve has a local minima denoted Jlim . This ux is the maximum
allowable ux loading if the settling is to be successful. If the in uent ux to the settler
6
is larger than Jlim , the sludge blanket will increase, resulting in solids (sludge) in the
euent.
For a general total ux model Jt (X ), the limiting ux can be obtained by solving
J (X ) = 0. To nd the minimum we have to check which of the extreme points that
t
dX 2
fulll JdX(X2 ) > 0. These calculations may have to be solved numerically. Obviously,
t

graphical solutions are also possible to use.

The limiting ux can be used for the design of the area of the settler. The ux at the
level where the in uent is located is
Jin = QAin Xin (20)
where Xin is the solid (sludge) concentration of the in uent water. We must require
that Jin < Jlim and hence
A > JQin Xin (21)
lim
Note that Jlim in general depends on A.
3.3 Numerical solution of the settler*
In order to model the settler during non-steady state conditions, the involved mass
balance have to be solved. This is done by applying the conservation law (continuity
equation) which states that the increase of mass per time unit equals the incoming ux
minus the outgoing ux. This can be stated as a partial dierential equation as
@X = , @J (22)
@t @z
where z is the height coordinate.
The most popular approach to solve (22) numerically is to divide the settling basin in
a number of horizontal slices (typically 10 to 100 slices). Each slice is regarded as well
mixed. A further description of numerical methods is, however, outside the scope of this
note.