Pre-Ozonation of Potable Water
Anna Möllergren
Department of Chemical Engineering 1, Lund Institute of Technology, Sweden and Hunter Water Australia
Abstract:
This study examines the suitability of pre-ozonation of potable water treatment at Grahamstown Water treatment
plant. Pilot plant experiments were performed to investigate the effectiveness of pre-ozonation on coagulation of
suspended colloids in the raw water. The removal of suspended particles was assessed through source water and
filter effluent measurements of turbidity. In particular focus were the positive effects of ozone in various
dosages, with and without the addition of alum, mainly by measuring turbidity, particle counts and colour. Other
tests such as the zeta potential, pH, temperature, algae count, UV absorption (254, 455, 665 nm) and
conductivity were measured for some of the samples. Following the examination of ozone influences on raw
water, the study was extended to investigate filter run times and head loss build-up. The jar test method and pilot
plant runs examined both conventional and direct filtration.
Experiments carried out on this pilot plant have shown that pre-ozonation lead to a decrease in the particle count,
colour and turbidity compared to the raw water. The ozone demand was 0.8 – 1.0 mg/L ozone.
Key words: Pre-ozonation; Ozone; Potable water; Destabilisation; Particle charge; Coagulation; Particle count
Introduction
All drinking water contains small concentrations of
impurities, such as organic and inorganic
compounds either in dissolved or in particular form.
Treating water so that it meets drinking water
standards for both appearance and safety can
require many processes – coagulation, flocculation,
sedimentation, filtration and residual disinfection.
Each process is used to achieve a particular goal
and any modification of a process to do double duty
can be useful to water utilities. Coagulation may be
such a process and therefore pre-ozonation of the
raw water prior to coagulation was investigated.
Ozone (O3) is a highly oxidizing molecule and has
been proven to effectively disinfect water without
creating dangerous disinfection by-products, which
are formed when disinfecting water with chlorine
[1]. Ozone may also have other positive influences
on water treatment, such as improve the
flocculation process by alternating with the particle
surface chemistry of particles present in the water.
Oxidation is the break down of organic compounds
into simpler molecules and complete oxidation
converts an organic compound to carbon dioxide
and water.
Most colloids (i.e. silica, kaolinite, aluminum)
found in water have a negative charge due to the
NOM (natural organic matter) sorption on the
particle surface [2]. Stabilisation of colloidal
particles in water is caused by this layer of
absorbed organic molecules, which makes them
repel one another and stay in dispersion due to
electrostatic repulsion [3]. To coagulate these fine
particles their electrical charge must be reduced or
neutralised.
Ozone affects NOM in two ways; (i) oxidative
cleavage of larger molecules and (ii) an increase in
polarity because of an increase in acidic functional
groups [2]. Ozonation may lead to desorption of
humic materials from the particle surface, as a
result of the increase in smaller molecules and the
more polar carboxylic groups. The organic layer
will decrease in thickness and the smaller, more
polar groups will be less of a steric hindrance. This
will allow the particles to come together and
aggregate more easily [2].
Materials
Chemicals used in this project were Alum and
Ozone. Dried and compressed air was used for
ozone production. The alum used at Grahamstown
was delivered as a liquid from a company called
Omega Chemicals and contains approximately 50
% active ingredients i.e. Aluminium sulphate.
Method
Equipment
An Ozgen WT-10C ozonator was used to generate
ozone gas by the “Corona discharge” method using
air. A Confined Plunging Liquid Jet Contactor
(CPLJC) was used to transfer the ozone gas into the
water phase, see figure 1. The CPLJC introduced a
jet of water through a nozzle into the cylindrical
downcomer by the suction effect created. This
generated fine bubbles, which travelled downwards
with the water. The downcomer outlet was
submerged in water in the riser, creating an airtight
chamber. Most of the mass transfer took place in
the mixing zone but also in the pipe flow zone [4].
Two different types of particle counters were used –
MetOne Model PCX Online Particle Counting
Sensors and a MetOne portable. Turbidity
1
measurements were made using Great Lakes online turbidity meters and a Hach 2100AN. Colour
was manually measured against de-ionised water
with the lowest detection limit of <5 Hazen Units
(HU) at Grahamstown WTP lab.
UV absorption (254, 455, 665 nm) and conductivity
were measured for some of the samples.
The impact of ozone on particle removal by settling
and filtration was investigated using the jar test
method at the pH of the raw water. The ozone
demand of the raw water was determined by
gradually increasing the ozone feed gas
concentration until maximal ozone production
occurred. The ozone demand is the point where a
residual is measured in the water, meaning that
everything that can be oxidized has been oxidized.
Conventional water treatment was simulated in
some pilot plant runs and direct filtration was also
simulated, by letting the flow leaving the
flocculation tank directly enter the sand filter. Only
one line was used for some runs to get familiar with
the equipment and to study the ozone induced
changes of the raw water. The parallel lines were
used when comparing the ozonated water and the
raw water before and after sand filtration and to
examine filter run times and head loss.
Results and Discussion
Jar Test
Pilot Plant Facility
The investigation was performed at a permanent,
two-line, pilot-scale facility located at Grahamstown
Water Treatment Plant at Grahamstown, Newcastle
(Australia). Each line of the pilot plant included a
rapid-mixing tank, a flocculation basin with three
compartments and a sand filter column. In the line
with the CPLJC, ozone gas was drawn into the
water pumped from the raw water tank and
collected in an ozonation tank before pumped
upstairs to a rapid mixing tank where alum was
injected by a dosing pump. The water then entered
the flocculation tank containing three cells, each
with an independently driven motor and paddle
with variable speed control to aid coagulation. The
outgoing stream then entered the bottom of a
clarifier and was drawn from below the water
surface to a batch wise sand filter. The effluent was
collected in a clear water storage tank and the
filtered water was reused for sand filter back
washing.
Optimum Ozone and Alum doses
Jar Test 2-5
No Ozone
2.5
Settled Turbidity
(10 min)
Figure 1: Schematic of a Confined Plunging Liquid Jet
Contactor [4]
Results from jar tests confirmed that ozone has an
effect on particle count and turbidity, especially
particle count. A decrease in the raw water colour
from 10-15 to <5 HU was observed and measured
after ozone doses of about 0.3 mg/L without the
addition of alum. Results presented from jar tests
on filtered turbidity are referring to water filtered
through a 0.45 µm paper filter.
It was observed during jar tests that, after the
addition of alum, floc formed faster in the unozonated water jars than in the ones with ozonated
water, especially for the jars with a higher alum
dose. However, as seen in figure 2, the ozonated
water resulted in lower settled turbidity than the unozonated samples for low alum doses (< 30 mg/L).
Ozone 0.5 mg/L
2
Ozone 1.0 mg/L
1.5
Ozone 1.4 mg/L
1
0.5
0
0
Experimental procedure
The experimental procedure consisted of jar testing,
ozone demand tests and pilot plant runs. In
particular focus for the investigation were the
positive effects of ozone in various dosages, with
and without alum, mainly by measuring turbidity
(NTU), particle counts and colour. Other tests such
as the zeta potential, pH, temperature, algae count,
10
20
30
40
50
60
Alum dose (mg/L)
Figure 2: Settled Turbidity, Various Ozone and Alum Doses
Filtered turbidity (see figure 3) indicated that a low
dose of ozone improved the filtered turbidity
compared to the raw water sample at alum doses
over 20 mg/L.
2
decrease in particle count compared to raw water
samples was for the particle sizes 7-15 µm.
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
No Ozone
Average Particle count, Raw & Ozonated (1.0-1.6 mg/L)
Ozone 0.5 mg/L
Ozone 1.0 mg/L
2000
Ozone 1.4 mg/L
Particle count/ml
Filtered Turbidity
Optimum Ozone and Alum doses
Jar test 2-5
Average
Ozonated
Average Raw
1500
26 % decrease
1000
0
10
20
30
40
50
60
Alum dose (mg/L)
500
43 % decrease
24 %
decrease
36 % decrease
44 % decrease
26 % decrease
0
Figure 3: Filtered Turbidity, Various Ozone and Alum Doses
2-3
Particle count on filtered raw and ozonated water is
consistently lower for the ozonated samples when
comparing water with the same alum doses (see
figure 4).
Particle count /ml
20 mg/L alum, 0.5 mg/L ozone
25 mg/L alum
200
25 mg/L alum, 0.5 mg/L ozone
150
30 mg/L alum
100
As seen in figure 7, the particle count of the sand
filtered water decreased soon after ozone
production was initiated and increased when ozone
production was stopped. Also observed in figure 7,
was the increase in particle count of the filtered
water in the end of the run. Runs were stopped and
filters were back-washed when turbidity and
particle breakthrough occurred.
30 mg/L alum, 0.5 mg/L ozone
50
0
2-3
3-5
5-7
7-10
10-15
Particle Count Filtered water
1.0 mg/L ozone, 30 mg/L alum
Run 6
>15
Particle size (micrometer)
300
Figure 4: Particle Count, Filtered Raw Water and 0.5 mg/L
Ozone with various Alum Doses
Particle count/ml
Ozone demand tests showed that the residual
concentration in the water increased with ozone
dose and the consumed ozone was 30-45 % of the
generated ozone. The ozone demand presented in
figure 5 was between 0.8-1.0 mg of O3 per litre of
Grahamstown raw water. The consumed ozone
stayed rather stable after an ozone residual was
measured in the water since everything that could
be oxidised in the water was oxidised.
Ozone Demand
0.00
0.00
0.31
0.83
1.00
1.08
1.18
1.21
1.37
1.43
250
200
1.0 mg/L ozone
150
100
50
0
0.5 1.5 2.5 3.5 4.5 5.5 6.5 7.5 8.5 9.5 10.5 11.5 12.5 13.5 14.5 15.5 16.5 17.5
Hours
Figure 7: Particle Count on Sand Filtered water with and
without ozone production
Consumed Ozone
Ozone Off gas
Ozone residual
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
2-3 micrometer
3-5
5-7
7-10
10-15
>15
Ozone production off
Ozone Demand
Ozone (mg/L)
>15
Pilot Plant Run
20 mg/L alum
250
5-7
7-10
10-15
Particle size (micrometer)
Figure 6: Average Particle Count on raw & Ozonated water
Particle count, raw & 0.5 mg/L ozone
Jar test 11
300
3-5
The decrease in turbidity due to ozone is harder to
determine than the decrease in particle count, as
seen in figure 8. Both runs (one line with ozone and
the other with raw water) had the same alum dose
and showed similar results in filtered turbidity.
However, soon after the start of the runs, it was
obvious that the head loss build-up was slower for
the ozonated run compared to the raw water. This
means that the filters will run for a longer time and
will need to be back-washed less frequently, which
is obviously desirable.
1.56
Ozone Dose
(
/L)
Figure 5: Ozone Demand measuring the ozone off-gas, residual
and consumed ozone
In figure 6 the average particle count is presented
for the ozone doses 1.0 – 1.6 mg/L. The largest
3
Turbidity F1, ozone
35 mg/L alum, 0.5 mg/L ozone
Run 12
•
Smaller size floc lead to a longer head loss
build-up time and a more even distribution
of the floc in the filter media.
•
An ozone dose between 0.5-1.0 mg/L
combined with an alum dose of at least 30
mg/L gave the best results over-all in
turbidity, filtered turbidity and particle
count.
Headloss F1, ozone
Turbidity F2, alum
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
Headloss F2, alum
ozone 0.5 mg/L
0
1
2
3
4
5
6
7
8
9
10
11
Time (hours)
Acknowledgments
Figure 8: Parallell runs with ozone and raw water
Results from UV absorbance measurements,
conductivity and the zeta potential did not result in
any conclusions of the ozone effect on particle
chemistry since there was little change from the raw
water results. The algae counts showed little change
of the ozone dosages 1.0 – 1.5 mg/L. Raw water
turbidity, particle count, pH, colour and
temperature stayed stable relatively stable
throughout the experiments.
Conclusions
The Pilot Plant study results cannot be summarised
with one ozone dose and one alum dose that would
give the best results. The effects of pre-ozonation
are likely to vary with the nature of source water
and other factors. Carefully controlled pilot studies
with parallel treatment lines are needed to further
determine the overall impact of pre-ozonation on
performance of the various stages of potable water
treatment. Results from this study are as following:
•
•
•
My supervisors, Prof. Geoffrey Evans at School of
Engineering, University of Newcastle, Craig
Jakubowski and Artur Majerowski at Hunter Water
Australia, and Prof. Anders Axelsson at the
department of Chemical engineering 1, Lund
Institute of technology are gratefully acknowledged
for their help, support and guidance. I would also
like to thank Peter Dennis and Darren Bailey at
Hunter Water Australia for the opportunity to write
my diploma thesis at Grahamstown Water
Treatment Plant. I extend my thanks to Dave
Kingsland for performing chemical analyses and
useful advice on various problems.
Abbreviations
CPLJC = Confined plunging liquid jet contactor
DBP = disinfection by-products
HU = Hazen Unit
NOM = natural organic matter
NTU = nephelometric turbidity units
An ozone dose over 0.3 mg/L of
Grahamstown water will decrease the raw
water particle count by at least 20 %
without the addition of alum.
UV = ultraviolet radiation
An ozone dose over 0.5 mg/L
Grahamstown water will decrease
apparent colour of the raw water from
15 down to <5 Hazen Units without
addition of alum.
[1] CAMEL, V. & BERMOND, A. (1998) The use
of ozone and associated oxidation processes in
drinking water. Water Res., 32:3208.
of
the
10the
For the more part of samples, an ozone
dose > 0.5 mg/L of Grahamstown water
will lower the raw water turbidity.
•
Turbidity measurements showed that
ozonation improved the settling of floc at
low alum doses (<30 mg/L).
•
The optimum ozone demand dose is 0.8 –
1.0 mg/l of Grahamstown water.
•
Ozonation of water with addition of alum
lead to smaller and also slower floc
formation compared to the floc formed in
raw water with only alum.
WTP = water treatment plant
References
[2] JEKEL, M.R. (1983) The benefits of ozonation
treatment prior to flocculation process. Ozone: Sci.
& Eng., 5:21-35.
[3] JEKEL, M.R. (1986) The stabilization of
dispersed mineral particles by adsorption of humic
substances. Water Res., 20:1543-1554.
[4] JAKUBOWSKI, C.A., EVANS, G.M.,
DENNIS, P. & ATKINSON, B.W. (2003) Ozone
Mass Transfer in a Confined Plunging Liquid Jet
Contactor. Ozone: Sci. & Eng., 25 (1), 1-12.
4