Wat. Res. Vol. 31, No. 5, pp. 991-996.

1997
~ Pergamon © 1997ElsevierScienceLtd. All rights reserved
Printed in Great Britain
PII:S0043-1354(96)00343-0 0043-1354/97 $17.00 + 0.00

REMOVAL OF IRON FROM POTABLE WATER USING A

TRICKLING FILTER

G. D I M I T R A K O S MICHALAKOS, J. M A R T I N E Z NIEVA, D. V. VAYENAS and

G. LYBERATOS *@
Department of Chemical Engineering, University of Patras and Institute of Chemical Engineering and
High Temperature Chemical Processes, P.O. Box 1414, GR- 26500 Patras, Greece

(Received October 1994; accepted in revisedform October 1996)

Abstract--A trickling filter for iron removal from potable water has been constructed and tested. Iron
removal was found to be caused by both biological and physico-chemical iron oxidation. The extent of
each oxidation type was assessed. In addition, iron concentration profiles were taken along the filter depth,
for both biological and physico-chemical iron oxidation, under the same operating conditions. Significant
quantities of iron precipitated, mandating a periodic filter backwash. The possible duration of the filter
operating cycles was studied as a function of the inlet iron concentration and the volumetric flow-rate.
The limits of feed iron concentration and hydraulic loadings on the filter were determined and appropriate
operating diagrams were constructed. © 1997 Elsevier Science Ltd

Key words---biofilm, biological oxidation, iron removal, physico-chemical oxidation, potable water,
trickling filter

I!NTRODUCTION at a pH under 6 and reduced forms may persist for

When iron is present in a water supply, at
concentrations exceeding the secondary drinking
water standards of 0.2mg/l (EEC, 1980), it is
undesirable for the following reasons (Kontari, 1988):
(a) iron precipitates give water a reddish colour when
exposed to air; (b) iron gives water a metallic taste;
(c) home softeners become clogged by iron precipi-
tates; (d) deposition of iron precipitates in the
distribution system can reduce the effective pipe
diameter and eventually clog the pipe; and (e) iron is
a substrate for growth of bacteria in the water mains.
When iron bacteria die and slough off, bad odours
and unpleasant tastes may be caused.
Surface water iron is generally found in its
precipitated ferric form. It is found in its ferrous state
in the deeper layers of some water reserves lacking
oxygen, or in groundwater, in which case it is in a
reduced dissolved form (Fe 2+) and often chelated.
The state of iron in water depends above all on the
pH and the redox potential. A dissolved form of iron
(FC ÷ or FeOH +) can be changed to a precipitated
form (FeCO3, Fe (OH)2 or Fe (OH)3) by increasing
either the oxidation potential, the pH, or both (Hem,
1961).
The rates of ferrous iron oxidation by air increase
with the pH and ;about 90% conversion may be
achieved in 10-20 rain at a pH of 7 (Sundstrom and
Klei, 1979). Iron oxidation is substantially slower
*Author to whom all ~-orrespondence should be addressed.
some time in aerated waters (Sawyer and McCarty,
1978). The rates may be increased by the presence
of certain inorganic catalysts or through the action
of micro-organisms. There are several genus of
bacteria that oxidize dissolved iron with different
mechanisms. Gallionella sp, Leptothrix ochracea and
Crenothrix polyspora cause primary intracellular
oxidation by enzymatic action, while secondary
extracellular oxidation is caused by the catalytic
action of polymers excreted by the filaments
comprising the stalks produced by Gallionella sp.,
the sheaths from Leptothrix sp., Crenothrix sp.,
Clonothrix sp. and Sphaerotilus sp. Also, secondary
extracellular oxidation is caused by the catalytic
action of polymers, excreted by the microscopic
filaments of capillary appearance that are aban-
doned by the gliding trichomes of Toxothrix
trichogenes and extracellular polymers that are
excreted by various Siderocapsaceae (Mouchet,
1992).
Aeration followed by solid-liquid separation is the
most commonly used physico-chemical treatment
method for iron removal. Aeration is generally
recommended for oxidizing ferrous iron for water
with high iron concentrations ( > 5 mg/l) so that
costs for chemicals can be avoided (Wong, 1984).
Sedimentation and/or filtration (sand or dual-media
filtration) are applied for the solid-liquid separation.
If necessary, other treatments may be added, such as
pH correction (lime softening), chemical oxidation
(with chlorine, potassium permanganate or ozone)

991
992 G. Dimitrakos Michalakos et al.
ion exchange processes, settling and biological
oxidation.
As mentioned above, numerous bacteria are
capable of provoking the biological oxidation of iron
inside or Outside their sheaths. The extent to which
an observed iron removal may be attributed to
physico-chemical and/or biological iron oxidation
may vary depending on the particular treatment
environment employed. In this communication, the
use of a pilot-scale trickling filter for iron removal
from drinking water is presented. The limits between
physico-chemical and biological iron removal are
identified. Experimental profiles of total iron
concentration are given in both cases (only physico-
chemical and combined physico-chemical and bio-
logical) and the limits of the filter for iron
concentration and hydraulic loadings were assessed
and used to generate an operating diagram, useful for
the design of appropriate trickling filters.
MATERIALS AND METHODS
The pilot-scale trickling filter consisted of a 1.8 m high
Plexiglas tube with an inside diameter of 9 cm. This
pilot-filter height is typical of a full scale industrial filter
(Metcalf and Eddy, 1991). Since it is the loadings (hydraulic
and iron) per unit cross-sectional area that count, no
scale-up is necessary. The filter was externally covered in
order to avoid algal growth on the walls. The support media
was gravel with a mean diameter of 0.5cm and a
~ 1330 m2/m3 specific surface area, while the depth of the
support media was 1.6 m and the filter porosity was 0.35
(Vayenas and Lyberatos, 1995, 1994). At the top of the filter,
there was a fixed nozzle which distributed the incoming
water evenly to the whole filter surface. An underdrain
system collected the treated water and any biological solids
that would detach from the media. Along the filter depth
there were 15 sampling ports for iron concentration
measurements in the bulk liquid. Thus it was possible to
have an experimental assessment of the total iron
concentration profiles along the filter depth (Figs 3-5).
Throughout all experiments the temperature was fairly
constant at about 28°C, the pH ranged from 7.0 to 7.3 and
the concentration of the dissolved oxygen in the liquid phase
was about 8 mg/l (saturation concentration). The above
measurements were made using the U-10 Water Quality
Checker (HORIBA). With these operating conditions and
since the redox potential ranged from 0.3 to 0.4V (as
measured with a redox potential electrode), according to
Hem (1961), we could have both physico-chemical and
biological iron oxidation taking place.
The use of a trickling filter has the advantage of not
requiring an external air supply, since air is naturally
convected through the filter due to the temperature
difference between the interior and exterior of the filter. A
similar filter referred to as a contact filter (filter media was
crushed rocks to 30-70 mm) has been mentioned by Carlson
and Schwertmann (1987). In our filter however, complete
aeration and very good filtration may be effected at the same
time, since the mean gravel diameter was only 5 ram.
Inoculum was taken from the wastewater treatment plant
of the University of Patras and added to the filter, with a
concentrated iron solution (FeSO4-7H20) which also
contained small quantities of ammonia (NH~CI) and
phosphorus (KH2PO,), to ensure bacterial growth. A
start-up time of approximately 10 d was necessary for the
biological oxidation to begin. During this period, the filter
was operated as a batch reactor. The filter was aerated
during start-up through the use of an air pump. After 10 d,
microscopic observations (x 1200) revealed the presence of
two types of microbes; cylindrically shaped (probably genus
Leptothrix and Crenothrix Cohn), and bean shaped
(probably genus Gallionella Ehrenberg). The filter was then
switched to continuous operation. During continuous
operation, water was added with feed iron concentrations
ranging between 1 and 17 mg/l. This covers the range of
possible concentrations encountered in drinking water
supplies. The feed solution resulted after mixing a
concentrated iron solution (containing small amounts of
ammonia and phosphorus) with water from the water
supply network (containing nonmeasurable iron amounts)
at a mixing chamber located at the top of the filter. The
concentrated iron solution was always kept in a refrigerator
located close to the filter, in order to prevent bacterial
growth in the feed medium.
In order to study the extent of physico-chemical iron
oxidation, the filter was disinfected, with commercial bleach
of chlorine (NaOCI 5%), to exclude bacterial growth. The
filter was disinfected every 2 d by adding chlorine to the filter
for half an hour and then passing network water through the
filter, before starting to feed with the iron concentrated
solution. Under these conditions, only physico-chemical
iron oxidation was possible.
During all experiments, pH, temperature and total iron
concentration were measured on a daily basis. Iron
concentration measurements were done according to the
Standard Methods for the Examination of Water and
Wastewater (APHA et al., 1989) using an atomic absorption
spectrophotometer (model SpectrAA-10 Plus, VARIAN).
EXPERIMENTAL RESULTS
The experiments aimed at determining the limits of
hydraulic and iron concentration loadings, under
which the filter efficiency is satisfactory (total iron
concentration at the filter outlet less than 0.2 mg/l).
The experiments also aimed at determining the extent
of biological vs physico-chemical iron oxidation
since, as mentioned earlier, both take place under the
operating conditions used.
Once continuous operation was established, the
volumetric flow rate (VFR) was progressively raised
to 1000 ml/min (corresponding to a retention time o f
3 min). At this V F R , various feed iron concentrations
were tested. Figure 1 presents the required duration
of operating cycles (time intervals in days, between
two successive backwashes of the filter) for total
iron concentrations in the range of 1-17 mg/1. At
1 mg/l, no backwash was required. It seems that a
steady-state was reached. For higher iron concen-
trations the duration of operating cycles had to be
decreased. At 17 mg/1, an operating cycle duration of
1 d was found sufficient, while the outlet iron
concentration was about 0.2 mg/l (limited operation).
Generally, as investigated, for higher concentrations
and VFRs, an increase in the inlet iron concentration
or in the V F R , decreases the required duration of the
operating cycles. During an operating cycle, only
minimal suspended solids or iron precipitates could
be observed at the filter outlet, while at the end o f
each cycle an increase of the suspended solids and
precipitates was apparent. This was used as a
criterion for starting filter backwash.
Once the feed iron concentration limit (17 mg/1) of
 Trickling filter for iron removal from water 993

20-

18-

16-

"0
v 14-
Z
0 12-
m
I--
<
ff-
10-
a
w
_J
o
>-
0

2-

0
0 4 6 8 1'0 1'2 1'4 l's 1'8 20
IRON CONCENTRATION (mg/I)
Fig. 1. Required backwash frequency (cycle duration) as a function of inlet iron concentration.

the filter for a VFR of 1000 ml/min was determined,
we searched for the feed concentration limits for
higher VFRs. Figure 2 presents the curve (upper
curve, which resulted after fitting the experimental
data) of iron oxidation. Above this curve the
operating cycles were of the order of only hours and
the filter outlet concentration was above 0.2 mg/1.
Below this curve, the filter had excellent performance.
We also found that the absolute upper VFR limit for
this filter was 3000 ml/min, since for higher VFRs the

20

-m-Biological and Physicochemical oxidation

1 El
-&- Physicochemical o x i d a t i o n
16

1 zl.

Z
o
n"
I-'- 1 0
Z
U..I
0
Z
0 8 -
0
Z
0 El x x
m----

,A

I I I
0
1000 1500 2000 2500 3000
VOLUMETRIC FLOW RATE (rnl/rnin)

Fig. 2. Operating curves which showing the limits when: (a) simultaneous biological and physico-chemical
iron oxidation take place; and (b) only physico-chemicaloxidation takes place.
994 G. Dimitrakos Michalakos et al.

filter flooded, regardless of what the feed iron
concentration was.
In order to assess the filter performance with
physico-chemical iron oxidation alone, the same
experiments were repeated after the filter was
disinfected. Iron oxidation was possible only
physico-chemically, using the air which flowed
physically through the filter. The lower curve of
Fig. 2 represents the physico-chemical iron oxidation
limiting curve, under which the filter had excellent
performance. It is obvious that the existence of
bacteria in the filter can dramatically improve the
filter efficiency under the same operating conditions.
Figures 3-5 present total iron concentration
profiles along the filter depth for both biological
and physico-chemical oxidation. As seen in Fig. 3
(inlet iron concentration: 4.1 mg/l, VFR: 2000 ml/
min) the biological oxidation profile is, as expected,
below the physico-chemical one, all along the filter
depth. We can also observe that the outlet iron
concentration with physico-chemical oxidation alone
is above the upper permitted limit of 0.2 mg/1 while,
with biological oxidation allowed, the outlet iron
concentration is less than 0.1 mg/l.
Figures 4 (inlet iron concentration: 4.0 mg/1, VFR:
2500ml/min) and 5 (inlet iron concentration:
3.0 rag/l, VFR: 3000 ml/min) show that differences in
the concentration profiles for the two types of
oxidation are more pronounced at the lower part of
the filter, suggesting that bacterial oxidation naturally
takes place in that part of the filter. Under these
operating conditions the flow rates were high and
resulted in a better exploitation of the total filter
height. The outlet iron concentrations for biological
oxidation still remained within the allowable limits,
whereas physico-chemical oxidation would be insuffi-
cient again. Also, comparison between Figs 3 and 4
shows that for the same inlet iron concentration the
biological activity is significant only for the lower
VFR (Fig. 3).
CONCLUSIONS
In this work the use of a trickling filter for iron
removal from drinking water was examined and
verified. The filter had excellent performance under
various flow rates and inlet iron concentrations. The
use of the trickling filter for iron removal confirmed
that natural air convection through the filter secures
oxygen saturation conditions at the bulk liquid. Thus,
there was no need for extra aeration. In addition,
since large quantities of iron precipitates were
remaining in the filter, oxidation and filtration were
probably taking place simultaneously within the same
treatment unit. The required frequency of filter
backwashing was found to depend on the inlet iron
concentration and the volumetric flow rate. The
higher the setting of these operating parameters were,
the more frequent was the required backwash.
Operating diagrams were constructed showing the
allowable limits in hydraulic loading and inlet iron
concentration so that the filter performs adequately.
Such limiting curves were given also for the case

Biological and Physicoehemical oxidation

4.5-
Physicochemical oxidation
4!
E
z 3.5-
O
3-
Ii
~-
Z
2.5-
LLI
O 2
Z
O l
1.5-
Z
O l
n- 1
m •
,&
0.5

0
0 2'0 6'0 go 1 ,0 16c
FILTER DEPTH (cm)
Fig. 3. Iron concentration profiles along the filter depth for: (a) simultaneous biological and
physico-chemical iron oxidation (inlet iron concentration: 4.1 mg/l, VFR: 2000ml/min); and (b) only
physico-chemical oxidation.
 Trickling filter for iron removal from water 995

Biological and Physiooehemical oxidation

4.5
Physi¢ochemical oxidation

z 3. -
0
W-
< 3-
n-
Z 2.5-
LU
CO 2-
Z
0 !1
co 1.5-
Z
0
n" 1

0.5
!
0
o 2o 4'0 6'0 8'o 160 1;~0 lz;O 160
FILTER DEPTH (cm)
Fig. 4. Iron concentration profiles along the filter depth for: (a) simultaneous biological and
physico-chemical iron oxidation (inlet iron concentration: 4.0 •g/l, VFR: 2500 ml/min); and (b) only
physico-chemical oxidation.

of physico-chemical oxidation alone. For the The advantages of biological iron removal of the
same VFRs, biological oxidation allows treatment presented trickling filter, are numerous: (a) rapid
of almost double the iron concentration than oxidation and no need for an external air supply;
physico-chemical oxidation alone. (b) iron oxidation and filtration taking place in the

5-
Biological and Physicochemical oxidation
4.5"
Physicochemical oxidation

O)
4-
E
Z
3.5-
O
31
l-
Z
2.5-
LU

Z 2-
O
1.5"
Z
Orr IB
1

0.5 L

o-
0 2'0 4'0 60 ab 160 1~0 1~,0 160
FILTER DEPTH (cm)
Fig. 5. Iron concentration profiles along the filter depth for: (a) simultaneous biological and
physico-chemical iron oxidation (inlet iron concentration: 3.0 mg/I, VFR: 3000 mi/min); and (b) only
physico.chemical oxidation.
996 G. Dimitrakos Michalakos et al.
same unit; (c) reagents are not required for pH
correction and flocculation; (d) high removal
capacity is attained since iron is retained in a very
compact form; (e) high filtration rate due to the
robustness of the biological floc; and (f) very
economical washing, since the amount of wash water
required is small enough compared to the amount of
treated water.
REFERENCES
APHA, AWWA and WPCF (1989) Standard Methods
for the Examination of Water and Wastewater, 17th
edition. American Public Health Association, American
Water Works Association and Water Pollution Control
Federation, Washington, D.C,
Carlson L. and Schwertmann (1987) Iron and manganese
oxides in Finnish ground water treatment plants. Water
Res. 21, 165-170.
Degremont (1991) Water Treatment Handbook, 6th edn.
Lavoisier, Paris.
EEC-Official Journal of the European communities,
no 80/779, August 30, 1980.
Hem J. D. (1961) Stability field diagrams as aids in iron
chemistry studies. J. A WWA, 53(2), 211.
Kontari N. (1988) Groundwater, iron and manganese:
an unwelcome trio. Water/Engng Mgmt, 25-26.
Metcalf and Eddy (1991) Wastewater Engineering. Treat-
ment-Disposal-Reuse. McGraw-Hill, Singapore.
M ouchet P. (1992) From conventional to biological removal
of Iron and Manganese in France. J. A WWA, 84(4),
158-166.
Sawyer C. N. and McCarty P. L. (1978) Chemistry for
Environmental Engineering, 3rd edn. McGraw-Hill,
Singapore.
Sundstrom and Klei (1979) Wastewater Treatment. Pren-
tice-Hall, Englewood Cliffs, N.J.
Vayenas D. V. and Lyberatos G. (1994) A novel model for
nitrifying trickling filters. Water Res. 28, 1275-1284.
Vayenas D. V. and Lyberatos G. (1995) On the design of
nitrifying trickling filters for potable water treatment.
Water Res. 29, 1079-1084.
Wong J. M. (1984) Chlorination-filtration for iron and
manganese removal. J. Am. Water Wks Assoc. 76, 76-79.