You are on page 1of 10

Valerie Seh Yi Ling, A0100802A

Water Filtration

ESE3401
Water and Wastewater
Engineering 1
Water Filtration Lab Report

Done by: Valerie Seh Yi Ling A0100802A


Date of Experiment: 8th Sept 2014

Valerie Seh Yi Ling, A0100802A


Water Filtration

1. Introduction
The objective of this experiment is to demonstrate the operation of the water
filtration process. In this experiment, we will determine the general hydraulic
characteristics and effect of variables such as the type and size of filter media on
the operation of the water filtration process. The type of filter is varied using fine,
coarse and mixed particles. We will then determine the type of filter media that
should be used for the treatment of raw water.
2. Theoretical Background
In the experiment, water will be flowing through the filter media of different
sizes. There will be some form of head loss across the filter bed due to resistance
from the flow. The experimental apparatus is designed in a way such that
particulates smaller than the passage between adjacent grains will be removed
and pores will be closed when accumulating deposits interstices downwards. As
such , particles that are entering the pores are transported deeper into the filter
bed and may escape into the effluent. This causes an increasing flow rate which
will increase the head loss across the filter bed.
We will be using the Bernoulli equation to determine the head loss for each filter
tube. Assuming that the total energy possessed by the flowing fluid is the same
at every cross section along the filter tube, Bernoulli equation is defined as:

P1
v 12 P 2
v 22
+ z 1+ = + z 2 + +hl
g
2 g g
2g
By rearranging the Bernoulli equation, we can get:

P1P2
v12v 22
hl=
+(z 1z 2)+
g
2g
where h is the head loss between P1 and P2
is the density of water : 1000kg/m3
(Z1 - Z2) is the depth of the filter media : 800mm
g is the gravitational acceleration : 9.81m/s 2
P1 and P2 is the pressure at the top and the bottom of the filter
In this experiment, V1 and V2 are assumed to be the same.
The second equation that we will be using is:

V=

Q
=
A

Q
D 2
( )
2

Where D is the internal diameter of flow filter tube : 172mm


Q is the flow rate = 0.8L/min (in part 2 of experiment)
2

Valerie Seh Yi Ling, A0100802A


Water Filtration

3. Experimental Procedure
The experimental set up is as shown in Fig 1 below. The experiment is carried out
using three types of filter media containing fine particles of different
composition. The type of filter media used are fine, coarse and mixed.

Picture 1: Pictures of Fine, Coarse and Mixed Sand Filters (From


left to right)

Part 1: Determining relationship between filtration rate and head loss

An inflow supply of water from a constant head tank will be directed to the filter
tubes. The water will be passing through the filter media in each tube. We will
then adjust the flow rate controller on the effluent line to vary the flow rates.
Pressure values at the inlet and outlet of the filter tube are recorded and
experiment is repeated at different flow rates. Head loss will be calculated
afterwards by using the pressure values obtained and the Bernoulli equation as
shown earlier in the report.
Part 2: Determining the effect of filter media on time length of filter runs
Turbid water is used in this experiment. We will use a turbidity meter to measure
the initial turbidity of the water supply. The flow rate of the turbid water through
the filter media will be adjusted to 800L/min. We will record the pressure values
at the inlet and outlet for each filter tube at every five minutes for 30 minutes.
For the filter tube containing the fine medium, pressure values will be recorded
every minute for the first five minutes as there might be a possibility of clogging
due to more particles in the turbid water. Two effluent samples will be collected
at the end of the 30 minutes filter run to measure its turbidity. Head loss will also
be calculated afterwards using the Bernoulli equation to aid the investigation of
effect of filter media on time length of filter runs.
At the end of the experiment, a back wash is conducted to clean the filter bed.
The flow of water is in the opposite direction, from the outlet, at a rate high

Valerie Seh Yi Ling, A0100802A


Water Filtration
enough to fluidize the bed to allow trapped residues to escape with the
overflowing wash water
4. Relationship between filtration rate and head loss
4.1 Experimental Results and Calculations

D 2
0.172 2
A= ( ) = (
) =0.0232m2
2
2

Using Bernoulli's equation:

P1P2
v12v 22
+(z 1z 2)+
g
2g

hl =

where g = 9.81m2/s,

= 1000kg/m3, (z1 - z2 ) = 0.8m, v12 = v22

Sample Calculation of head loss, where Q=0.024m 3/hr for fine sand,

P1P2
v 12v 22 19.114.6
hl=
+ ( z 1z 2 )+
=
+ ( 0.8 ) +0=0.341 m
g
2g
(1000 )( 9.81)

Flowrate
Q
(m3/hr)

Fine Sand
P1
h
(kPa P2
(m)
)
(kPa)

Filtration
rate, V
(m2/hr)

0.024

1.034

14.6

19.1

0.036

1.552

14.4

17.3

0.048

2.069

14.1

15.2

0.06

2.586

13.7

13.2

0.072
3.103 13.1 9.4
Table 1. Tabulation of Flow rate,
media
Log V
Fine
0.015
0.191
0.316

-0.467
-0.297
-0.162

Coarse Sand
h
P1
P2
(m)
(kPa) (kPa)

0.34
1
0.50
4
0.68
8
0.85
1
1.17
7
Pressure

Log h
Coarse
-2.310
-2.310

14.2

22.2

14.0

21.8

13.6

21.4

13.2

20.9

12.4
19.9
readings, and

Mixed
-0.831
-0.702
-0.621
4

Mixed Sand
P1
h
(kPa P2
(m)
)
(kPa)

0.01 14.
5 7
21.1
0.148
0.00 14.
5 5
20.4
0.199
0.00 14.
5 1
19.6
0.239
0.01 13.
5 7
18.8
0.280
0.03 13.
5 0
17.3
0.362
Head loss through filter

Valerie Seh Yi Ling, A0100802A


Water Filtration
0.413
-0.070
-1.821
-0.553
0.492
0.071
-1.450
-0.442
Table 2. Tabulated values for log V and log h

Graph of Log h vs Log V


0.500
Fine

0.000
f(x) = 1.09x - 0.5
0.000
0.200
0.400
-0.500
R = 0.99
f(x) = 0.78x - 0.85
Log h -1.000
R = 0.98

0.600

Mixed
Linear (Mixed)
Coarse

-1.500
-2.000

Linear (Fine)

f(x) = 2.95x - 3.01


R = 0.84

Linear (Coarse)

-2.500
Log V

Graph 1. Plot of Log h vs Log V


Derivation of an equation relating head loss and filtration rate: Log h = a LogV +
log b
Rearranging the equation, Log h = Log bVa . Hence, h= bVa
h: head loss (m), V: filtration rate (m/hr), a & b are constants
a
b
h
Fine Media
1.092
10-0.4966= 0.319
0.319V1.092
-3.0129
Coarse Media
2.948
10
=
0.000971V2.
948
0.000971
-0.851
Mixed Media
0.776
10
= 0.141
0.141V0.776
Table 3. Tabulated results from Graph 1
4.2 Discussion
From graph 1 and the equation H= bV2 derived above, we can deduce a
relationship whereby the filtration rate (V) is directly proportional to the head
loss (h). As seen from the graph, an increase in filtration rate leads to an increase
in the head loss for all three filter medium. However, we notice that the rate of
change of head loss with respect to filtration rate varies across the different filter
mediums. Based on the gradient of the graph, which is also the a value tabulated
above, we can deduce that the fine sand filter is the most affected by filtration
rate.
This deduction can be supported in the change in head loss values recorded
during the experiment, whereby the fine sand filter has the highest head loss
ranging from 0.341 to 1.177m. The mixed sand filter medium is the next most
affected by filtration rate with a head loss range of 0.148m to 0.362m. Lastly, the

Valerie Seh Yi Ling, A0100802A


Water Filtration
coarse sand filter has the lowest head loss ranging from -0.015m to 0.035m,
showing that it is least affected by filtration rate.
For this experiment, we assume that the flow velocities and the depth of the
filter media remains constant throughout. This means that the pressure drop
across the filter media would be the only variable. Head loss is the reduction in
the total head or pressure of a particular fluid when it moves through a fluid
system and in this case, from the top to the bottom of the filter media. The pipe
is filled with fine, coarse and mixed sand filter. As such, when there is a
downward flow of water from the tank, it will result in a build up of resistance
between the filter media and the water, leading to a loss in energy. Thus, we can
deduce that the increase infiltration rate will increase the resistance which
eventually leads to a higher head loss.
5. Effect of filter media on time length of filter runs
5.1 Experimental Results and Calculations
For the second part of the experiment, the flow rate Q is adjusted to 0.8L/min.
The pressure readings are taken at 5 min internal for 30 minutes. As for the fine
sand medium, additional readings are taken at 1 min interval for the first 5
minutes for precaution in the event clogging takes place.

V=
The filtration rate would be:

Removal efficiency =

Q ( 0.8 x 0.001 )60


=
=
2
A
0.172
(
)
2

2.066 m/hr

T f T i
100 , where T: Turbidity
Tf

Sample Calculation for Fine medium: Removal efficiency=

4000.894
x100%=
400

99.78%
Turbidity (f)
Average
Sample 1
Sample 2
0.894
Fine
0.921
0.866
2.120
400 NTU
Coarse
2.18
2.06
1.260
Mixed
1.44
1.08
Table 4. Initial and Final turbidity of samples after treatment
Filter
Medium

time (hr)

Turbidity
(i)

P1
(kPa)

Fine Sand
P2
(kPa)

0.000

14.2

0.017

14.1

Removal Efficiency
(%)
99.78
99.47
99.69

Coarse Sand
Mixed Sand
h
P1
P2
P1
P2
(m)
(kPa)
(kPa)
h (m)
(kPa)
(kPa)
h (m)
0.66
15.5
7
13.5
21.4 -0.005
14.2
19.6
0.25
0.66
15.4
7
6

Valerie Seh Yi Ling, A0100802A


Water Filtration
0.033

14.1

15.4

0.050

14.1

15.5

0.067

14.1

15.5

0.083

14.1

15.3

0.167

14.1

15.3

0.250

14

15.1

0.333

13.9

15

0.417

13.8

14.9

0.500
13.7
14.8
Table 5. Pressure and head
across time

0.66
7
0.65
7
0.65
7
0.67
8
13.5
21.4 -0.005
14.1
19.6 0.239
0.67
8
13.5
21.3 0.005
14.1
19.6 0.239
0.68
8
13.4
21.3 -0.005
14.1
19.6 0.239
0.68
8
13.4
21.2 0.005
14.1
19.6 0.239
0.68
8
13.4
21.2 0.005
14.1
19.6 0.239
0.68
8
13.3
21.2 -0.005
14.1
19.6 0.239
loss readings for fine, coarse and mixed filter media

Log T
Fine
-1.778
-0.176
-1.477
-0.176
-1.301
-0.182
-1.176
-0.182
-1.079
-0.169
-0.778
-0.169
-0.602
-0.162
-0.477
-0.162
-0.380
-0.162
Table 6. Tabulated Log T and log h values

Log H
Coarse
-2.301
-2.301
-2.301

Graph of Log h vs Log T

Log h

0
-2 -1.8-1.6-1.4-1.2
-1
-0.8-0.6-0.4-0.2
f(x) = 0.01x - 0.16
-0.5
R = 0.66
f(x) = - 0.62
-1
R = 0
-1.5
-2

Log T

f(x) =-2.5
- 2.3
R = 0

Graph 2. Plot of Log h vs Log T


7

Fine
Linear (Fine)
Coarse
Linear (Coarse)
Mixed
Linear (Mixed)

Mixed
-0.622
-0.622
-0.622
-0.622
-0.622

Valerie Seh Yi Ling, A0100802A


Water Filtration
Derivation of an equation relating head loss and filtration rate: Log h = n LogT +
Log K
Rearranging the equation, Log h = Log KTn . Hence, h= KVn
h: head loss (m), V: filtration rate (m/hr), n & K are constants
n
K
h
Fine Media
0.0141
10-0.157= 0.697
0.697V0.0141
-15
-2.301
Coarse Media
1x e
10
= 0.005
0.005V e-15
Mixed Media
1x e-14
10-0.622 = 0.239
0.239V e-14
Table 3. Tabulated results from Graph 1
5.2 Discussion
Based on graph 2 that is plotted, the number of points plotted for each graph are
different. This is especially for the coarse and mixed media, whereby negative
head loss are experienced in some of the data points and no calculations can be
made. This will affect the accuracy of the graph. We will cover this error in the
later section of the report.
From graph 2, the R2 value for all three graphs are low, with the fine filter
medium graph having a R2 value of 0.6648, the coarse filter medium of 2E -15 and
the mixed filter medium with a negligible R2 value. We can conclude that the
power law relationship is not an accurate representation of the relationship
between head loss and the time of filter run. Despite the fact that no relationship
can be drawn from the graph plotted, we can use theoretical knowledge and
experimental results to predict the possible relationship head loss and time.
Based on experimental results as seen on table 6, we can observe that the head
loss increases as time increases, even though the increment is extremely
minimal. For example, for the fine sand medium, there is only a 0.001m increase
in head loss over the half hour period. This could be attributed to the
accumulation of particles over time as they get clogged in between the small
sand particles with smaller pore size which can increase the resistance build up.
As compared to larger particles such as a coarse filter medium, the coarse
particles have relatively larger gaps so the resistance build up is more likely to
be slower, and the lead loss would be lesser. This is supported by the
experimental results whereby the head loss for coarse filter medium is lesser
than that for fine filter medium. Hence, we can conclude that the fine sand
particles will experience a greatest initial head loss with an increment
afterwards.
We can also use the percentage removal of turbidity to identify the effectiveness
of the media used for sand filtration. From table 4 above, we can see that all
three filter mediums have a high level of removal efficiency based on the
turbidity values. The fine sand filter has the highest capability of removing
particles, as it has the highest removal efficiency of 99.78%. This is followed by
the mixed sand filter of removal efficiency 99.69% and lastly, the coarse sand
filter of 99.47%. Based on pure theoretical results alone, fine sand filter would be
recommended for the treatment of raw water.

Valerie Seh Yi Ling, A0100802A


Water Filtration
However, we must take into account that while the fine sand particles are
smaller, packed more compactly and able to trap more colloidal particles, it
would require a high frequency of back washing . In addition, the time taken for
backwashing would be longer as sand particles, having a lower mass, would take
a longer time to settle down. A This would mean that the cost incurred for the
treatment of raw water using a fine sand filter would be more.
The coarse sand filter has a larger particle size. This means that it will also have
a higher porosity and lesser resistance, resulting in a lower head loss. However,
by having more small gaps between its filter media and higher porosity, lesser
colloidal particles would be trapped and the output water quality would be less
desirable. This is reflected in the turbidity values, where the removal efficiency of
the particles is the lowest at 99.47%
The mixed sand filter would be recommended instead as it not only has the
second highest removal efficiency of 99.69%, the top portion of the mixed filter
media is filled with activated carbon. This allows adsorption of most organic
compounds which can be present in raw water originating from lakes and rivers.
This can contribute to the reduction in turbidity and removal of particles. The
frequency of backwashing needed would also be lesser as compared to fine filter
medium, allowing the cost incurred to be lesser. Hence, taking the cost and
removal efficiency into consideration, it would be a practical choice to
recommend a mixed filter medium as it has a strong advantage over the other
two filter mediums.
At the end of the experiment, backwashing of all three filters were carried out.
Water at high pressure is pumped upwards to expand and unclog the particles
trapped in between the filter media. The backwashing flow rate must be
controlled in a manner such that it is sufficiently high enough to fluidise the bed
to flush out particles, yet not excessively high to prevent unnecessary loss of
sand particles. With reference to PUB, the backwash interval for sand particles is
typically 24 to 36 hours for it to be effective. As the experiment is only carried
out over a period of half hour, this could explain why the head loss observed is
minimal as a longer period might be needed for the head loss to be observable.
During the experiment, there are some significant errors that lead to negative
head loss for the calculations. These possible errors are included below:
1. There might be parallax error. As the flow meter is located at a relative low
place, it is difficult to take and adjust the reading at eye level. This would
lead to an inaccuracy of the data. One way to improve this accuracy is to
lie down on the floor such that the reading can be taken at eye level.
2. The equipment used are faulty. During the conduction of the experiment,
water can be seen leaking from the equipment, resulting in a drop in water
pressure. The multiple negative head loss results obtained from the
experiment could be attributed to this. For experimental results to be more
accurate, a check could be done with the lab technician in charge to see if
any adjustments could be done to the apparatus to fix the fault.
9

Valerie Seh Yi Ling, A0100802A


Water Filtration
3. The presence of air bubbles may affect turbidity results. As turbidity meter
measures light deflection of suspended particles in the sample tube, it is
important that we ensure that there are no air bubbles present when
placed into the turbidity meter. We should tap the sample tube slightly to
ensure that it is free of air bubbles to ensure that readings taken are
accurate.
4. Turbidity readings not taken immediately after sample tube is placed
inside the turbidity meter may result in lower turbidity readings. This
happens when the particles inside the sample tube starts to settle down.
As such, turbidity readings should be taken right immediately after the
sample tube is placed inside the turbidity meter.
5. Fluctuations in the reading of the flow mater may affect accuracy of
results. As the flow meter readings take a while to stabilise, the time
allocated during the lab session may be insufficient for the readings to
stabilise. As such, the values obtained may be less accurate. This could be
rectified by repeating the experiment a few more times. An average could
be taken after to determine the flow meter reading.
6. Conclusion
We can conclude from the results of the experiment that the filtration rate and
head loss are directly proportional to each other. As the filtration rate increases,
the head loss will increase as well. This is attributed to the increase in resistance
between the particles when the flow rate increases. We can also conclude that
the particle size affects head loss, where smaller particles such as fine sand has
the highest head loss.
In the selection of the filter media for the treatment of raw water, other than
considering the removal efficiency of the type of filter media used, we must also
take into consideration other factors such as cost. In the case of this experiment,
the performance of the fine media is better than coarse and mixed media
medium. However, taking into account the high frequency of backwash that
needs to be conducted for fine medium, it is worthy to note that the cost of
backwashing outweigh the effectiveness in this case as the difference in turbidity
removed between the fine and mixed media is not very significant. As such, I
would recommend using the mixed filter medium as the most viable option for
the treatment of raw water.
7. References
Engineers Edge. (2014). Fluid Flow Head Loss Darcy- Weisback Equation.
Retrieved from: http://www.engineersedge.com/fluid_flow/head_loss.htm
National University of Singapore. (2014). ESE3401 Water and Wastewater
Engineering Lab Manual.
PUB. (2012, March). Innovation in Water Singapore. Retrieved from:
https://www.ewi.sg/pdf/publication/InnovationWater_vol2.pdf

10