Settling Tank Examples

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Numerical Analysis of Type I Settling

Saurabh Arun Samdani1 and Saurabh Kapoor 2

Abstract: Sedimentation and clarification are used interchangeably for potable water; both refer to the separating of solid
material from water. Various types of sedimentation exist, based on characteristics of particles. In type 1 settling, particles
whose size, shape, and specific gravity do not change over time are suspended in dilute solution In dilute solutions number of
particles is insufficient to cause displacement of water (most potable water sources). Terminal settling velocity can be found
out theoretically from the equilibrium of a particle settling through a column and used to calculate the removal efficiency.
However many parameters are not easy to measure and therefore empirical analysis of type I sedimentation is carried out
to determine the efficiency of removal for a particular loading rate. A mathematical model is formulated and a numerical
solution technique have been used for calculating removal efficiency from the time and concentration data of settling analysis.

Keywords: Numerical analysis, Sedimentation, Type I Settling

INTRODUCTION
Sedimentation and clarification are used interchangeably Examples:
for potable water; and both refer to the separating of solid
material from water (Peavy et al. 1985). Since most solids • Plain settling of surface water prior to treatment by
have a specific gravity greater than 1, gravity settling is rapid sand filtration (type 1)
used remove suspended particles. When specific gravity is • Settling of coagulated and flocculated waters (type 2)
less than 1, floatation is normally used. Various types of • Settling of coagulated and flocculated waters in lime-
sedimentation exist, based on characteristics of particles soda softening (type 2)
• Settling of waters treated for iron and manganese con-
Discrete or type 1 settling particles are those particles tent (type 1)
whose size, shape, and specific gravity do not change
over time. PARTICLE SETTLING
Particle settling, or sedimentation, may be described for
Flocculating particles or type 2 settling particles are a singular particle by the Newton’s equation for terminal
those particles that change size, shape and perhaps spe- settling velocity of a spherical particle. A knowledge of this
cific gravity over time. velocity is basic in the design and performance of a sedimen-
Above types have both dilute and concentrated suspensions tation basin. The rate at which discrete particles will settle
in a fluid of constant temperature is given by the Stokes
Dilute: In dilute suspension, number of particles is insuffi- equation for laminar flow:
cient to cause displacement of water (most potable wa-
ter sources). g(ρp − ρw )d2
vt = (1)
18µ
Concentrated: In concentrated suspension, number of
particles is such that water is displaced (most waste where
waters) g = accelaration due to gravity
ρp = density of the particle
Sedimentation has many applications in prepara- ρw = density of the water
tion of potable water as it can remove suspended d = diameter of particle
solids and dissolved solids that are precipitated. µ = coefficient of viscosity
However direct application of above equation is not seldom
1
Undergraduate Student, Department of Civil Engineering, possible in water treatment because the size of particles d
Indian Institute of Technology Guwahati, Guwahati-781039, mus t be known and the correction factor φ to account for
India. email:samdani@iitg.ernet.in. departure from sphercity has to be determined. An indi-
2
Undergraduate Student, Department of Civil Engineering, rect method of measuring velocities of discrete particles in
Indian Institute of Technology Guwahati, Guwahati-781039, dilute susupensions and of determining setling characteris-
India. email:skapoor@iitg.ernet.in. tics of a suspension, was devised by T.R.Camp 1943( cited

1/ Samdani and Kapoor


from (Peavy et al. 1985)). Settling column is constructed 2. article with diameter dp < d0 will have a settling
as shown in Fig.1 Suspension is placed in column and com- vp < v0 , will arrive at or pass the sampling port in
pletely mixed and then allowed to settle quiescently. time t0 , provided its position is below Zp .
3. If the suspension is uniformly mixed, i.e. particles are
randomly distributed,. then the fraction of particles
with size dp having settling velocity vp which will ar-
rive at or pass the sampling port in time t0 will be
Zp /Z0 = vp /v0 . Thus the removal efficiency of any
size particle from suspension is the ratio of the set-
tling velocity of that particle to the settling velocity
v0 defined as Z0 /t0 . These principles can be used to
determine the settleability of any given suspension,
using shown apparatus. Theoretically depth is not a
factor but for practical reasons, 2 m is usually chosen.

PROCEDURE

1. Determine C0 of completely mixed suspension at time


zero.
2. Measure C1 at time t1 . All particles comprising
C1 have a settling velocity less than Z0 /t1 , where
v1 = Zo /t1 . Thus, the mass fraction of particles re-
moved with v1 < Zo /t1 is given by x1 = C1 /C0 .
3. Repeat process with several times ti , with the mass
fraction of particles being vi < Z0 /ti
4. Values are then plotted on a graph to obtain Figure
2, where the fraction of particles remaining for any
settling velocity can be determined
5. For any detention time t0 , an overall percent removal
(X ) can be obtained. That is, all particles having
a settling velocity greater that v0 = Z0 /t0 , will be
removed 100% (un-hatched area in Figure 2 ). Thus
FIG. 1. Settling column for analyzing type 1 suspension 1 − x0 particles are completely removed in time t0
The remaining particles have a vi < v0 (hatched area
in Figure 2), and will be removed according to ratio
Particle placed at the surface has an average settling ve- vi /v0 .
locity of: 6. If the equation relating v and x are known, then the
distance travelled Z0 area can be found through integration using Eq. 4
v0 = = (2)
time of travel t0 Z x0
vi
X = 1 − x0 + dx (4)
Another particle placed at distance Zp , terminal velocity less 0 v0
than the surface particle, but arrives at the same time, has
a settling velocity of: where X is the total mass fration removed by sedi-
Zp mentation.
vp = (3)
t0
NUMERICAL SOLUTION
which is less than v0 ; but arrives at sampling port a same Eq. 4 is usually solved graphically by plotting xi vs vi
time. Thus, the travel time for both particles is equal, where and then finding the area under the curve till limits of in-
t0 = Z0 /v0 = Zp /vp and vp /v0 = Zp /Z0 = h/H Thus tegration. The method is tedious and prone to errors. The
some generalized statements can be made concerning the area under the curve can be found out numerically. How-
above relationships. ever since the point v0 upto which area is to be determined
is generally not an element of the set of available (calculated
1. All particles having a diameter equal to or greater ) points, the curve has to be interpolated smoothly upto v0
than d0 , i.e. have settling velocity greater than v0 , to find the value of x0 . Interpolation formulas (Scarborough
will arrive at or pass the sampling port in time t0 . 1966) are available for

2/ Samdani and Kapoor


• Start of the interval – Newtons’s forward difference 1.0
interpolation formula.
• End of the interval – Newtons’s backward difference

Fraction of Particles with Velocity


interpolation formula.
r
• Middle of the interval – Gauss’s central difference in- o

Less than vo; ri = Ci/Co


terpolation formula, Stirling’s formula, Bessel’s for-
mula. w
c
dr
However above formulas are valid only for equal intervals. In i
the problem at hand the intervals of xi and vi are not uni-
form, so none of the above formulas could be used. Among
the interpolation formulas for unequal intervals, Lagrange’s
and Newtons’ general interpolation formulas are different
forms of the same equation with lagrange’s formula best
suited for computer implementation. 0
type-1.pre

In Lagrange’s formula, values of f (x) at x0 , x1 , x2 , . . . , xn Z o


points are known and denoted as y0 , y1 , y2 , . . . , yn . The aim Settling Velocity, vt =
is to find a value of y for a particular value of x. Lagrange’s t
formula for y is given by
FIG. 2. Removal efficiency as a function of settling time
(y − y1 )(y − y2 )(y − y3 ) . . . (y − yn )
y = x0 (5)
(y0 − y1 )(y0 − y2 )(y0 − y3 ) . . . (y0 − yn ) Comparison of original points and
0.7
points interpolated from Lagrange’s formula
Interpolated points
(y − y0 )(y − y2 )(y − y3 ) . . . (y − yn ) Original points
+x1
(y1 − y0 )(y1 − y2 )(y1 − y3 ) . . . (y1 − yn ) 0.6
Mass fraction remaining

(y − y0 )(y − y1 )(y − y3 ) . . . (y − yn ) 0.5


+x2
(y2 − y0 )(y2 − y1 )(y2 − y3 ) . . . (y2 − yn )
0.4
+...
(y − y0 )(y − y1 )(y − y2 )(y − y3 ) . . . (y − yn−1 ) 0.3
+xn
(yn − y0 )(yn − y1 )(yn − y2 ) . . . (yn − yn−1 )
0.2

An implementation of the lagrange’s formula was tried but


the interpolated points are undulating as shown in Fig.3. 0.1

Lagrange’s formula works well for the upper portion where


0
the original points and the interpolated points overlap, but -0.01 -0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035

in the lower portion, the graph is an undulating one with Velocity in m/min
some of the values even negative which is not physically FIG. 3. Interpolated points from Lagrange’s formula
possible.
In the numerical interpolation of this point (x0 , v0 ), there
are many hurdles and many standard methods cannot be removal efficiency X is calculated for all the available values
used because: of vi Xi denotes the removal efficiency for v0 = vi .
Px0
vi · ∆x
1. Intervals are unequal for vi or xi so solution cannot X = 1 − x0 + xi =0 (6)
be obtained by usual integration techniques like simp- v0
son’s rule, weddle’s rule (Scarborough 1966). Further calculations are done on the basis of the problem
2. Lagrange interpolation is undulating. type.
3. To use splines, first and second derivatives at the start
Type a: v0 is given
and end of the interval is required. Hence in this case
In this case value of X is required to be found out for given
splines also could not be used.
v0 . From the stored values Xi , value of X is interpolated as
The only option available was to use linear interpolation Xi+1 − Xi
between two data points. Correspondingly integration is X = Xi + (v0 − vi ) (7)
vi+1 − vi
performed using Trapezoidal rule(Scarborough 1966) which
is applicable to any number of subintervals, whether even such that
or odd and for all interval sizes. Using the trapezoidal rule vi+1 ≥ v0 ≥ vi (8)

3/ Samdani and Kapoor


Type b: X is given TABLE 1. Calculations for ex1.inp.
In this case value of v0 is required to be found out for given
X. From the stored values vi , value of v0 is interpolated as Time (min) Ci ( mg/L) xi Velocity vi
0 650 1.00 inf
vi+1 − vi 58 560 0.86 0.034483
v0 = vi + (X − Xi ) (9)
Xi+1 − Xi 77 415 0.64 0.025974
91 325 0.50 0.021978
such that 114 215 0.33 0.017544
Xi ≥ X ≥ Xi+1 (10) 154 130 0.20 0.012987
250 52 0.08 0.008000
ALGORITHM AND FLOWCHART
The flowchart for the algorithm is shown in Fig.7 Program
recieves the input in form of a input file from which all the
required information is extracted. The next step involve the 60 189
calculation of velocities and the mass fraction. Once the 80 180
velocities and mass fractions at various instants are known 100 168
removal efficiencies at all these instants are calculated using 130 156
equation 4. After deciding the problem type, case specific 200 111
interpolation is done to get the desired result in any of the 240 78
two cases. 420 27

IMPLEMENTATION For analysing the data from this input file, following com-
The algorithm was coded in the C programming language mand is used at the prompt ($)
on a Linux based platform. For graph plotting Gnuplot soft- $ settlingtypeI exmpl
ware has been used and the output of the program is a pdf In this case exmpl is the name of the input file. The output
file processed by LATEX. To run the program, an input file is of the program is found in the file exmpl.inp.report.pdf.
required. In settling analysis two kinds of problems can be
defined CASE STUDIES
1. When the loading rate (v0 ) is known and removal Example 1
efficiency is required to be found out. In this case A Settling column analysis is run on a Type-
option X is used in the input file. 1 suspension. The settling column is 2 m tall.
2. When the removal efficiency (X) is known and cor- The initial concentration of well mixed sample is
responding loading rate (v0 )is required to be found 650 mg/l. Result of the analysis are shown below
out. In this case option v is used in the input file. Time, min 58 77 91 114 154
Conc., mg/L 560 415 325 215 130
A sample input file to be used with the program is shown.
All numerical entries are typed after the colon. Input Units: What is the theoretical efficiency in a settling basin with
Loading rate v0 in m / min a loading rate of 2.4 x 10−2 m3 /m2 d(m/d)? The input file
Removal Efficiency X in decimals ( not %) for this problem is ex1.inp
Concentration, Ci in mg/ L
Time, ti in min Problem type ( X or v):X
Height of column,H in m Value of X if v or v0 if X:0.024
Height of the column:2
No of data points:7
Type your entries after the colon(ie :) : 0 650
If you want to calculate X 58 560
with v given take option X : 77 415
If you want to calculate v0 91 325
with X given take option v: 114 215
Your option Problem type ( X or v): X 154 130
Value of v if X or X if v:0.0174 250 52
Height of the column: 2
No of data points:8 After analysing this data with settlinganalysisI pro-
Time concentration: gram, the efficiency X is found to be 78.49 % for given load-
0 300 ing rate v0 = 0.024000 m/ min

4/ Samdani and Kapoor


Calculations are shown in Table 1 and the graph of vi vs.
Cumulative settling velocity distribution curve
xi is shown in Fig. 4. In Fig. 4, it can be seen that the ver- calculated from the settling velocity data given in Table 1
tical line of v0 cuts the curve at an intermediate point. The
0.9
area of the curve is found out only upto this point. Please v0 = 0.022982
0.8
note that in the table the entries are given in increasing or-
der of time, while calculations are done in increasing order 0.7

Mass fraction x
of vi , with the assumption that at t = ∞ , vi = 0 and xi = 0. 0.6
0.5

Cumulative settling velocity distribution curve 0.4


calculated from the settling velocity data given in Table 1 0.3
0.9 0.2
v0 = 0.024000
0.8 0.1
0.7 0
0 0.005 0.01 0.015 0.02 0.025 0.03
Mass fraction x

0.6
0.5 Velocity v in m/min

0.4 FIG. 5. Graph for inv.inp


0.3
0.2
What is the removal efficiency in a settling basin with a
0.1
loading rate of 25 m3/m2*d (m/d)? The input file for this
0 problem is ex2.inp
0 0.005 0.01 0.015 0.02 0.025 0.03
Velocity v in m/min Problem type ( X or v):x
FIG. 4. Graph for ex1.inp Value of X if X or v0 if v:0.0174
Height of the column: 2
No of data points:8
Another possibility is with a given value of X the loading Time concentration:
rate is required to be determined. For the same data if one 0 300
wishes to find the value of loading rate for efficiency 80 %, 60 189
the input file will be 80 180
100 168
Problem type ( X or v):v 130 156
Value of X if v or v0 if X:0.80 200 111
Height of the column:2 240 78
No of data points:7 420 27
0 650
58 560 An analysis was run using this input file and the the effi-
77 415 ciency X is 72.58 % for velocity v0 = 0.017400 m/ min The
91 325 In the inverse problem X is assumed to be 65 % and the
114 215 input file is prepared as
154 130
Problem type ( X or v ):v
250 52
Value of X if X or v0 if v:0.65
After analysing this data, for the given efficiency X = 80(%) Height of the column: 2
the loading rate v0 is found to be 0.022982 m/ min. The plot No of data points:8
for this input file is shown in Fig.5. Note that the value of Time concentration:
v0 is different than in the previous case. 0 300
60 189
Example 2 80 180
Settling analysis is run on a Type-1 sus- 100 168
pension in a typical 2 m column. The ini- 130 156
tial concentration is 300 mg/l. Data as follows. 200 111
Time, min 60 80 100 130 200 240 420 240 78
Conc, mg/L 189 180 168 156 111 78 27 420 27

5/ Samdani and Kapoor


TABLE 2. Calculations for ex2.inp. rate on the efficiency and the designing of a sedimentation
tank for a particular removal efficiency.
Time (min) Ci ( mg/L) xi Velocity vi
0 300 1.00 inf REFERENCES
60 189 0.63 0.033333 Peavy, H., Rowe, D., and Tchobanoglous, G. (1985). En-
80 180 0.60 0.025000 vironmental Engineering. McGraw Hill International Edi-
100 168 0.56 0.020000 tions.
130 156 0.52 0.015385 Scarborough, J. (1966). Numerical Mathematical Analysis.
200 111 0.37 0.010000 Oxford and IBM Publishing Co.
240 78 0.26 0.008333 T.R.Camp (1943). “Velocity gradients and internal work in
420 27 0.09 0.004762 fluid motion.” Boston Society of Civil Engineering, 30(3),
219–230.

Again, analysis was run using this input file and the velocity
v0 is obtained as 0.023494 m/ min for given efficiency X of
65.00 ( % ).
Calculations are shown in Table 2 and the graph of vi vs.
xi is shown in Fig. 6.

Cumulative settling velocity distribution curve


calculated from the settling velocity data given in Table 2
0.7
v0 = 0.017400
0.6
0.5
Mass fraction x

0.4
0.3
0.2
0.1
0
0 0.005 0.01 0.015 0.02 0.025 0.03
Velocity v in m/min
FIG. 6. Graph for ex2.inp

CONCLUSIONS
Removal efficiency is a critical parameter in the design of
sedimentation tank. However calculating removal efficieny
manually is a tedious job that is best left to the computer.
The data obtained from settling analysis experiment is the
concentration of solids in various samples taken at different
times. Even though the samples may be taken at equal inter-
vals, the main variables in finding the removal efficiency are
velocity and mass fraction left. Since the interval in velocity
and mass fraction are unequal, many popular numerical in-
terpolation and integration methods are not applicable and
the simple method of trapezoidal rule has to be used. The
algorithm is implemented on a Linux based machine and
the output file contains the plot of mass fraction vs. veloc-
ity. This software can be used study the variation of loading

6/ Samdani and Kapoor


START

Input

Calculate Velocities

Calculate Mass fraction

Calculate Efficiency
Tables

v Problem X
Type?
v or X

Search for i such that Search for i such that


Xi+1 < X < Xi vi < v0 < vi+1

Interpolate v0 from Interpolate X from


corresponding vi & vi+1 corresponding Xi & Xi+1

Plot the graph &


display result

STOP

FIG. 7. Flowchart for Type 1 settling analysis

7/ Samdani and Kapoor


Output file of Settling analsis of type–1 sedimentation for input file
: inv.inp
Saurabh Arun Samdani, Saurabh Kapoor
B.Tech Fourth year,
Department of Civil Engineering,
Indian Institute of Technology Guwahati

The Efficiency X is 80.00 ( % ) for velocity v0 = 0.022982 m/ min

Cumulative settling velocity distribution curve


calculated from the settling velocity data given in Table 1
0.9
v0 = 0.022982
0.8
0.7
Mass fraction x

0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.005 0.01 0.015 0.02 0.025 0.03
Velocity v in m/min
Figure 1: Cumulative settling velocity distribution curve calculated from the settling velocity data given in
Table 1

Table 1: Data from a Type I sedimentation analysis as per inv.inp.


Time (min) Ci ( mg/L) xi Velocity vi
0 650 1.00 inf
58 560 0.86 0.034483
77 415 0.64 0.025974
91 325 0.50 0.021978
114 215 0.33 0.017544
154 130 0.20 0.012987
250 52 0.08 0.008000

You might also like