Aquacultural Engineering 21 (2000) 181 – 199

www.elsevier.nl/locate/aqua-online

Efficiency of nitrification in trickling filters

using different filter media
Odd-Ivar Lekang *, Helge Kleppe
Department of Agricultural Engineering, Uni6ersity of Agriculture, pb 5065, 1430 Aas, Norway

Received 2 February 1999; accepted 6 October 1999

Abstract
The efficiency of nitrifying trickling filters varies with media type. Three sizes of crushed
Leca (dried expanded clay) and three types of plastic media (Kaldnes rings, Norton rings,
and a rolled mat of Finturf artificial grass) were tested in columns 1.5 m high and 11 cm in
diameter. Synthetic water containing 1.5 mg total ammonia nitrogen (TAN) 1 − 1 and 1 mg
PO34 − l − 1 was added at a rate of 0.5 l min − 1 and a temperature of 15°C. The nitrification
rate stabilised 7–8 weeks after start-up in the Leca fractions and artificial grass columns. For
the other two media, no positive stabilization was observed over the duration of the
experiment (140 days). At the end of the experimental period, the nitrification rate was 100%
in the Leca fraction columns, 80% in the Kaldnes ring columns, 60% in the Norton ring
columns and 40% in the artificial grass columns. The nitrification rate was between 0.1 and
0.2 g TAN m − 2 day − 1 in the different columns, and the hydraulic load was below 0.25 m3
m − 2 day − 1. © 2000 Elsevier Science B.V. All rights reserved.

Keywords: Biofilter; Trickling filter; Nitrification; Filter media

1. Introduction

Various types of biological filters are commonly used in recirculating aquaculture

filters (Muir and Roberts, 1982; Van Rijn, 1996; Skjølstrup et al., 1997). A simple
and well-established design, which is also used in municipal wastewater (Metcalf
and Eddy, 1991; Rosenthal, 1993; Ferchini et al., 1994), is the trickling filter (TF).
Here the water flows down through a stationary filter media by gravity. The
advantages of trickling filters include: low maintenance, cheap installation, and

* Corresponding author. Tel.: +47-64-948784; fax: + 47-64-948810.

E-mail address: [email protected] (O.-I. Lekang)

0144-8609/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 1 4 4 - 8 6 0 9 ( 9 9 ) 0 0 0 3 2 - 1
182 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199

great tolerance of differences in hydraulic and organic loads. Trickling filters also
encourage oxygenation and removal of carbon dioxide from the water, important
functions in aquaculture. Several factors are important in the function of trickling
filters. Most important is to choose the correct filter media. The following factors
are important for selecting filter media in trickling filters:

1.1. Void ratio

The volume that remains filled with air after the media has been filled into the
filter housing divided by the total filter volume (Timmons and Losordo, 1997). Low
void ratio results in a longer contact time between the trickling water and the
established biofilm. The problems encountered with small void ratios include easier
clogging, lower transport of air from the surface to the lower part of the filter and
reduced water flow through the filter.

1.2. Specific surface area

It is important that the filter media have a large surface area, where biofilm can
grow. A larger area allows more bacteria to grow per unit volume of filter medium,
and results in increased removal of ammonia per unit of filter volume. It is also
cheaper to build a trickling filter using media with large surface area because it
occupies less space.

1.3. Weight

Filter media with low weight are preferable because they are easier to handle.

1.4. Homogeneous waterflow

To avoid dead zones and channels that will reduce the nitrification rate, it is
important to have a filter medium that gives the most homogeneous water flow
possible.

1.5. Economics

The filter media must have a reasonable price.

The filter media used in trickling filters can be classified by their construction, for
example granulated media and fixed media. Granulated media are normally small
separate elements, while fixed media consist of larger constructions with connected
areas. Granulated media can be separated into rock, sand, Leca (dried expanded
clay) and plastic element types.
Since the filter medium is of such great importance for the function of biofilters,
several experiments have been made both to study and to compare different filter
media types (Krüner and Rosenthal, 1983). Experiences with Leca in a submerged
upflow biofilter showed increased nitrification rate with a higher degree of crushing
 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199 183

(Wien, 1994a,b). It was also found that light materials have a higher nitrification
rate than heavy materials with the same size and degree of crushing (Wien, 1995).
The aim here was to compare the nitrification of six different filter media used in
trickling filters. Three types of Leca and three types of plastic media were
compared. It should be noted that for several of the filter media types used, there
is currently very little knowledge available of their use in trickling filters.

2. Materials and methods

The experiment was carried out over the period 28 July–15 December 1997. Six
media of different shape, weight and specific surface area were filled into columns
through which water containing ammonia trickled. The following media were tested
(Fig. 1):

2.1. Finturf artificial grass

A mat of Finturf artificial grass was rolled up, raised vertically and put into a
pipe (Fig. 1). The surface of the Finturf functioned as a trickling filter. The mat
measured 1.4× 0.45 m giving a total area of 0.63 m2. The specific surface area of
the mat was 284 m2 m − 3 and the density 138 kg m − 3 (Selseth, 1996).

2.2. Kaldnes rings

Kaldnes rings are small plastic rings in the form of pipes with a cross inside and
fins on the outside.
They have a diameter of 9.1 mm and a length of 7.2 mm and are specially
designed to for use in nitrification and denitrification filters. Their density is close
to that of water, 0.96 kg l − 1 (Hem et al., 1994; Rusten et al., 1995) and the specific
surface are was 500 m2 m − 3 (KMT, 1996).

Fig. 1. The media used in the experiment: Leca 2 – 4, 2 – 7, 4 – 10 mm, Kaldnes rings, Norton rings and
Finturf artificial grass.
184 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199

2.3. Norton rings

Norton rings are also plastic elements formed as pipes with crosses inside. The
diameter of the rings used in our experiments was 25 mm, and they had a density
of 88 kg m − 3 and a specific surface area of 220 m2 m − 3 (Sterner, 1997).

2.4. Leca

Leca is a product name of a light weight clay aggregate (LWA), which is a clay
product that is dried, expanded and formed into small balls under industrial
conditions. Leca can be delivered in spheres or crushed. In our experiments three
types of crushed Leca were tested with particle sizes 2–7, 4–10, and 2–4 mm. The
particle size 2 – 7 mm was a special type designed for use as a filter material and
treated with dolomite. The other fractions were ordinary Leca with no surface
treatment. There have been few measurements made on the surface area of Leca.
With the BET method (using gas adsorption in multimolecular layers, Brunauer et
al., 1938), specific surface areas between 100 000 and 1 000 000 m2 m − 3 have been
measured. Since only a small part of the trickling water passed through the
material, the exterior specific surface area (measured to be between 500 and 1000
m2 m − 3) is a better measurement (Wien, 1994a,b, 1995). The 2–7 mm Leca particle
sizes were also inoculated with Tetra-Bactozym capsules, which contain natural
organic material produced from aquatic plants. The purpose of this colloidal
material was to increase the activity of the nitrifying bacteria in aquariums. Eight
capsules were added to each column.
An experimental rig was constructed as follows. The rig contained a stand, a level
tank, distribution pipes, trickling columns, and outward channel (Fig. 2). The level
tank was placed on a higher level than the trickling columns so that gravity could
be used for water transport. Ordinary tap water flowed continuously into the level
tank. A heater and a thermostat controlled the water temperature at 15°C.
In the level tank, ammonia, and later phosphate, was added to the incoming
water to ensure development of nitrifying bacteria. The concentration of ammo-
nium was set to 1.5 mg l − 1 (136 g m − 2 of column surface per day), which
corresponded to 0.02 mg ammonia at the temperature and pH used. This is above
the recommended maximum values in aquaculture of 0.01 mg NH3 l − 1 (SECL,
1983; Knoph, 1995), requiring that the water be treated before reaching a fish tank.
This is also much lower than normal values in municipal wastewater that average
around 15 mg NH4-N l − 1 in Norway (FAKTA, 1995). The phosphate concentra-
tion in the inlet water was set to 1 mg PO34 − l − 1. The addition of phosphate was
not made from the start of the experiment because initially it was thought that the
phosphate levels in the inlet water were sufficient for biofilm development.
Nordeidet et al. (1993) have shown that a concentration greater than 0.15 mg l − 1
is required for optimal nitrification measured in a biorotor. Addition of phosphate
was made from September 17th and through the rest of the experiment. To ensure
good mixing of the added chemicals in the level tank, an air pump was used.
 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199 185

Fig. 2. Experimental system used for the experiment.

A solution containing 4.45 mg ammonium chloride (NH4Cl) l − 1 and 6.02 ×

10 − 4 ml phosphoric acid (H3PO4) was pre-mixed in a tank, and continuously added
to the level tank by a peristaltic pump to obtain the right concentrations of
ammonia and phosphate in the inlet water to the columns.
From the level tank, two 50-mm PVC pipes distributed the water over each of the
trickling filter columns. From the distribution pipes, 4-mm tubes led the water to
each column. The water flow to each column was kept stable at 0.5 l min − 1
throughout the experiment. This represented a hydraulic load of 91 m3 m − 2 top
surface area per day and is over the recommended minimum hydraulic loading for
biofilter media (Grady and Lim, 1980; Roberts, 1985).
The trickling columns consisted of 110 mm PVC pipes with a length of 150 cm.
Inside each pipe were two gratings, one at the bottom to prevent the biofilter media
from falling out of the column, and one at the top to ensure good distribution of
the water into the column. In each column was approximately 13 l of filter media.
There were 21 columns in total, three for each media.
The experiment was designed as a factorial with three replicates on each of the
six different media tested. The total experimental period lasted for 140 days. The
statistical analyses included analyses of variance (ANOVA) and Bonferroni tests
(5% level) executed in the program SYSTAT.
In the period July 28 – August 18, 1997 water samples were taken for analysing
twice a week. During the rest of the experiment, sampling was once a week because
only small changes were observed. The samples were taken at the same time each
day, between 18:00 and 21:00 h. Samples were taken of the inlet water to the
columns (i.e. the level tank) and of the outlet from each of the 21 columns. The
186 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199

water samples were analysed for pH, dissolved oxygen, temperature, ammonium
(total ammonia nitrogen, TAN), nitrite, and nitrate. At the end of the experimental
period (December 10th) the retention time of the water in the columns was
measured by adding a coloured tracer to the water.
Dissolved oxygen and temperature were measured with YSI model 59. pH was
measured by a Metrohm 744 pH meter. The nitrogen species were measured by a
HACH DR/200 spectrophotometer. A tablet of Rhodamine WT was dissolved in
30 l of water and added to the top of each column to measure the retention time
visually.

3. Results

3.1. Measured 6alues

3.1.1. Total ammonia nitrogen (TAN)

The average TAN concentration in the outlet water was relatively stable from all
the columns for the first 28 days. After this the concentration decreased in the
outlet water from all the media. In the columns filled with different Leca fractions,
the outlet concentration of TAN was relative stable after week seven, with almost

Fig. 3. The average ammonia concentration in the outlet water measured over the experimental period,
from trickling filter columns filled with different filter media: Leca, Kaldnes, Norton, Finturf. The value
for the inlet water to the column is also shown.
 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199 187

Fig. 4. The average nitrite concentration in the outlet water measured over the experimental period, from
trickling filter columns filled with different filter media: Leca, Kaldnes, Norton, Finturf. The value for
the inlet water to the column is also shown.

no TAN left (Fig. 3). The last three measurements taken with the plastic Finturf
columns were also stable. In addition the Norton rings showed a tendency for
stabilisation in the last 2 – 3 measurements. With the Kaldnes rings the TAN
concentration continued to decrease up to the end of the experimental period. The
average inlet concentration of ammonium was 1.3 mg TAN l − 1 (9 0.2).

3.1.2. Nitrite concentration

After 20 – 50 days there was an increase in the nitrite concentration, which
reached a maximum value after 48 – 70 days depending on the media (Fig. 4). The
concentration of nitrite reached a maximum first for the Leca media. The peak
value was also higher (about 1 mg NO2-N l − 1) with these media. After reading
their maximum values, the concentrations decreased to a low level in all columns
(B 0.1 mg NO2-N l − 1).

3.1.3. Nitrate concentration

After 40 – 60 days there was an increase in the nitrate concentration in the outlet
water from all the columns (Fig. 5). After 60–80 days the concentration was
relative stable throughout the rest of the experimental period. The concentration of
nitrate was highest in the outlet water from the columns with Leca, and lowest in
188 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199

the water from the columns filled with Finturf grass. The average inlet concentra-
tion was 2.1 mg NO3-N l − 1 ( 9 0.2).

3.1.4. Oxygen
The average oxygen concentration in the inlet water was 10.7 mg O2 l − 1 (9 0.3).
A decrease was observed in the oxygen concentration for all the trickling filter
columns through to the end of the experimental period. The greatest decrease was
measured in the columns filled with Leca, where the lowest value was measured
after 70 days at 6.7 mg O2 l − 1 ( 90.2). The plastic media also showed some
decrease. At the end of the experiment, the decrease in the oxygen concentration
was not so large. This is attributed to a small increase in the oxygen concentration
of the inlet water.

3.1.5. pH
Over the first 7 weeks, the pH values at the outlet of the trickling columns
showed little variability. After 7 weeks the value gradually fell for the remainder of
the period, for all of the media (Fig. 6). This fall occurred more rapidly for the Leca
columns than for the plastic media, and the decrease was also greater. In the outlet
water from the Leca columns, pH fell below pH 6.5. The pH in the inlet water was

Fig. 5. The average nitrate concentration in the outlet water measured over the experimental period from
trickling filter columns filled with different filter media: Leca, Kaldnes, Norton, Finturf. The value for
the inlet water to the column is also shown.
 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199 189

Fig. 6. The development in the average pH value in the outlet water, measured over the experimental
period, from trickling filter columns filled with different filter media: Leca, Kaldnes, Norton, Finturf.
The value for the inlet water to the column is also shown.

stable until day 98, with an average of 7.7 mg O2 l − l (9 0.1), before it subse-
quently fell.

3.1.6. Temperature
Until day 98 the average water temperature was 14.79 0.7°C. However, be-
cause of unexpectedly low winter temperatures, the heater did not manage to
raise the water temperature to 15°C. The last two measurements of water tem-
perature at the inlet showed a drop to 12°C (Fig. 7)

3.1.7. Retention time

The retention time of the water in the different columns varied considerably
over the experimental period (P B0.001) (Table 1). The longest retention time
(298 s) was measured in the columns filled with the smallest Leca particle size
(2–4 mm). The shortest retention time (9 s) was measured in the columns filled
with Finturf artificial grass. These correspond to an average water flow through
the columns of between 4.7 and 155.6 cm s − 1.
190 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199

3.2. Calculated 6alues

3.2.1. Total ammonia nitrogen

By calculating the difference in the amount of TAN in the inlet and the outlet
water, an expression of the amount of TAN nitrified in the trickling filter can be
found. For the columns filled with Leca, the consumption of TAN increased until
day 56, and then stabilized at an average value of 1.3 mg TAN 1 − 1 (9 0.2). This
was : 100% nitrification of the TAN in the inlet water (Table 2). All the plastic
media showed a lower degree of nitrification (PB 0.001) with the best values being
obtained for the Kaldnes rings with about 80% nitrification. The lowest nitrification
values of 36% were obtained in the columns filled with Finturf grass. Table 2 shows
that the nitrification rate was highest for the columns filled with Norton rings (0.2
mg TAN m − 2 day − 1), and lowest for the columns filled with Leca (0.1 TAN m − 2
day − 1). Looking at the nitrification per retention time unit in the trickling column,
by far the best values were obtained in the columns filled with Finturf artificial
grass, while the lowest values were obtained in the Leca columns.

Fig. 7. Inlet water temperature recorded over the experimental period.

Table 1
The retention time of the water measured 3 days before the end of the experiment (10.12.97), in
trickling filter columns filled with different mediaa

Column Retention time (s) Water flow (cm s−1)

Leca, 2–7 mm filter type (La) 179 9 10a 7.8 90.4

Leca, 4–10 mm ordinary type (Lb) 124 96b 11.3 9 0.6
Leca, 2–4 mm ordinary (Lc) 217 915a 6.5 90.4
Leca,2–4 mm inoculated ordinary(Ld) 298 919 4.7 90.3
Kaldnes rings (K) 72 926bc 19.4 9 7.0
Norton rings (N) 21 94cd 66.7 911.0
Finturf artificial grass (M) 9 9 3d 155.6 923.1

a
The water flow velocities through the columns are also shown. Mean values with different
superscripts differ significantly (a,b,c, PB0.05).
O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199 191
192 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199

3.2.2. Nitrite consumption

After 63 days all of the nitrite produced was oxidized to nitrate in the columns
filled with Leca (Table 2). At the end of the experiment, there were significant
differences in the concentration of nitrite remaining in the columns (PB 0.001),
with the greatest concentration in those filled with plastic media. The lowest values
were obtained in the columns filled with Kaldnes rings, and highest in the columns
filled with Norton rings and Finturf grass. The highest maximum production of
nitrite was obtained in the columns filled with Leca and Kaldnes rings. The highest
value overall was obtained in the outlet water from the trickling column filled with
inoculated Leca 2 – 4 mm (0.930 – 0.179) mg NO2-N l − 1).

3.2.3. Nitrate production

There were significant differences in the production of nitrate between the
different columns by the end of the experimental period (PB0.001) (Table 2). The
nitrate production was highest in both the Leca columns and the Kaldnes columns
with 1 mg NO3-N l − 1. The lowest production was obtained in the column filled
with Finturf grass.

3.2.4. Oxygen consumption

From the start of the experiment until day 56, the oxygen consumption in the
columns filled with Leca increased rapidly. No corresponding development was
seen in the columns filled with plastic media where oxygen consumption increased
slowly. On the final experimental day (140), there was a decrease in the oxygen
consumption in all the columns. Looking at the oxygen consumption in the last two
measuring days in the experiment (day 133 and 140), the highest oxygen consump-
tion were observed in the columns filled with Leca fractions with the smallest
particle sizes (2 – 4 and 2 – 7 mm). The consumption varied around 3 mg O2 1 − 1
(Table 2) (P B0.001). One of the Leca fractions (4–10 mm), as well as the plastic
media, showed an oxygen consumption around 1 mg O2 l − 1. Also, the oxygen
consumption per gram converted TAN was highest for the smallest Leca fractions
(Table 2).

4. Discussion

4.1. Experimental design

There were some problems with the experimental rig used. The small tubes
leading water out to the different columns were exposed to fouling and needed to
be regularly calibrated twice a week. However, between calibrations a maximal
divergence up to 9 70 ml min − 1 from the ideal 500 ml min − 1 was observed. At the
end of the experimental period there was a tendency for clogging to occur in the
trickling columns filled with Leca 2 –4 mm, both ordinary and inoculated. This
shows that the hydraulic capacity for this media had been reached.
 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199 193

The use of synthetic water could be questioned. The normal problems with
suspended solids and high levels of COD were not tested. In earlier work, an
equivalent experiment was made by taking inlet water from the outlet of the
University fish farm (Ask, 1996). Because the loading of fish and growth of fish in
the farm varied, there were great variances in the content of the outlet water, which
affected the main object of the experiment. For this reason synthetic water was
used.
The rapid increase in the production of nitrite, especially in the Leca columns,
after 4–7 weeks was probably a result of the addition of phosphate. This indicates
that the concentration of phosphorous, at 0.2 mg PO34 − l − 1, was too low and
restricted the development of biofilm. Increasing the concentration to 1 mg PO34 −
l − 1 (0.33 mg PO4-P l − 1) led to an increase in biofilm production. This corresponds
to experiments by Nordeidet et al. (1993) (rotating biodisk), which had a lower limit
of 0.15 mg PO34 − P l − 1.

4.2. Oxygen consumption

The plastic media used have a larger void ratio than most of the Leca columns,
and therefore the need for taking oxygen from the water is reduced. This corre-
sponds well with the results from the experiments. The Leca columns with the
largest particle size (4 – 10 mm) have a consumption that is more equal to that
obtained with the plastic media. This must be a result of the larger particle size.
Looking at the retention times for the different media, the Leca filters have
somewhat longer retention times. This shows that the conditions are close to
flooded, and can also help to explain why the oxygen consumption from the water
is higher. The maximum hydraulic load was reached for the Leca columns, and this
further demonstrates the flooded conditions in these columns.

4.3. TAN

The TAN consumption varied significantly between the different biofilter

columns, with the best results obtained for the columns filled with Leca. Here the
bacteria consumed \95% of the incoming TAN. This is a very high degree of
purification compared to earlier experiments. Also, for the columns filled with
Kaldnes rings a high degree of nitrification (80%) was obtained. Earlier experiments
comparing different media (Kaiser and Wheaton, 1983; Klemetson and Cohen,
1984; Rogers and Klemetson, 1985) showed a maximum nitrification of 35–60% for
trickling filters with a hydraulic load in the range 0.008–0.035 m3 m − 1 day − 1. The
nitrification rates calculated for the columns with plastic media (Finturf and
Norton) at the end of the experimental period lie close to these values. Columns
filled with Leca had a higher consumption of TAN. The reason is probably due to
the larger surface area of the Leca clay and the longer retention time in the
columns. A larger surface results in more contact between the biofilm and the
water, and diffusion into the biofilm is affected by the contact time.
194 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199

There were large differences in the retention time between the different Leca
media, but no effect on nitrification was seen. Experiments with trickling filters
show that the thickness of the biofilm, and the nitrification rate, decrease down the
column (Boller et al., 1994; Nijhof, 1995). The reason for this is the decrease in the
TAN concentration down the column. The columns filled with Leca 2–4 mm
(normal and inoculated), showed the longest retention times and are over-sized.
These columns could have been reduced in length and have still reached 100%
nitrification.
Because the columns with a large surface utilize much of the TAN in their upper
parts, low nitrification is obtained in the lower part of the filter. This is the reason
for the low total nitrification rate for the Leca columns. With increased TAN
loading into the columns, the nitrification rate would be higher for the Leca
columns, and it would have been possible to separate the nitrification rate between
the Leca columns. Under our conditions there was no benefit in using Leca with a
small particle size.
When evaluating filter media, the ability to avoid clogging is of interest since this
affects the hydraulic load of the filter. By the end of the experiment, two columns
filled with the smallest Leca particle size, 2–4 mm, showed a tendency to clog and
were operating at their maximum hydraulic load. In the columns with plastic media
no such problem was encountered, showing that a higher hydraulic load is possible.
If a larger load is desired using small Leca particles periodic back-flushing may be
a solution. The function would then be more equivalent to a submerged filter. In
practical aquacultural use with higher COD and SS loads, the maximum hydraulic
load for the Leca filter will be even more reduced. Normally these factors will
reduce the nitrification (Bovendeur et al., 1990; Hem et al., 1994; van Loosdrecht et
al., 1995). The experiment shows that for a given specific area, and disregarding the
possibility of clogging, Leca gives the highest nitrification rate.
From day 56 and to the end of the experiment, there were larger variations in
TAN consumption in the columns filled with plastic media than the ones filled with
Leca. This shows that the Leca columns are more stable filter systems. The
explanation for this is probably that the retention time in the columns is longer,
because of reduced void ratio. Clogging and opening of large pores in the biofilters
results in a larger variation in the flow.

4.4. Nitrite production and use

The maximum nitrite concentration in the outlet water from the different
columns varied as expected. The highest maximum values were found in the Leca
columns filled with 2 – 4 mm innoculated. The maximum values obtained for all the
media were above the lethal limit for fish (Liao and Mayo, 1974; Russo et al.,
1974). Clearly no fish could be in the systems during start-up. With all the media,
most of the nitrite was further oxidized to nitrate, and there was either no or a low
accumulation of nitrite in the system or none at all. For the plastic media, and
especially the Finturf mat, there was some accumulation of nitrite, but values were
below those recommended for aquaculture. Van Rijn and Rivera (1990) have
 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199 195

shown that nitrite accumulation can be a problem in water reuse systems involving
biological filters.

4.5. Nitrate production

All the media showed a decrease in the amount of oxidized TAN after a peak
between days 50 and 60. This corresponds to a peak in the nitrate production. The
explanation for this is probably due to a rapid increase in the number of nitrifying
bacteria followed by a fall as the competition between bacteria increases. After this
the number of bacteria stabilizes at a lower level. No obvious peak was obtained in
the columns filled with Norton rings and Finturf grass. This was probably a result
of the short retention time, which prevented a rapid increase in bacterial numbers.
In all the Leca columns there was low variation in the production of nitrate, whilst
more variation was seen among the plastic media. This may arise from the larger
void ratio and shorter retention time with the plastic media.

4.6. Start up period

The TAN consumption in the columns filled with Leca and Finturf grass
stabilized after 56 and 63 days, respectively. During this period most of the nitrite
produced was oxidised to nitrate. The concentration of nitrate remains stable. This
fact, and the stable oxygen consumption, indicates a well-established biofilm.
Comparing inoculated and uninoculated Leca no obvious differences in the start
up period were observed. A minor effect could be that the maximum nitrite
concentration was observed in the inoculated columns. However, there were no
significant data to show that the inoculation material, Tetra Bactozym, has any
effect on increasing the rate of biofilm creation. Also, Malone and Manthe (1985)
have tried commercial bacterial additives with no significant effect on the start up
time. A possible explanation of our results is that the inoculating material was
washed out of the columns before an effect could be obtained. The powder was
only added to the surface of the filter. If trickling filters are to be inoculated, the
retention time between the inoculation material and the biofilter media must be
increased.
For the columns filled with Norton rings, there was a tendency for the consump-
tion of TAN and oxygen to stabilize by the end of the experimental period.
However, an increase in the production of nitrate was still observed, therefore full
nitrifying was not achieved. In the columns filled with Kaldnes rings an increase in
the consumption of TAN was observed until the end of the experiment. This
indicates that biofilm was still being established after 140 days.
In earlier experiments, the time for establishment of biofilm was between 4 and
8 weeks (Hirayama, 1974; Bovendeur, 1989). A trickling filter experiment with
gravel as the medium, showed stable nitrification after 5 weeks (Vayenas et al.,
1997). This corresponds well with our results, with the exception of the columns
filled with Kaldnes and Norton rings. With these media there was no stabilization
after 20 weeks. An initial lack of phosphate in the inlet water may have caused a
delay in the start up time for the filters in our experiment.
196 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199

The biofilm in the columns filled with Finturf artificial grass stabilized more
quickly than that in the other plastic media. This may be accounted for by a higher
shear force generated by the water flowing through the Finturf media. This high
shear force kept the biofilm thin, thus enabling it to stabilize more quickly. The
longer start up times encountered with the plastic rings, compared with the Finturf
grass, also indicate that the geometrical shapes of the rings may cause clogging.
In all the columns the pH was seen to fall from 7.7 to 7.0. Boller et al., (1994)
found that such a drop in pH could reduce the nitrification rate to 1/4. of its
maximum value. By adding a buffer (HCO− 3 ) the nitrification rate could be
increased and, hence, the start up period decreased.

4.7. E6aluation of the nitrification rate

Earlier experiments have shown that the nitrification rates for trickling filters in
aquaculture lie between 0.01 and 0.32 g TAN m − 2 day − 1 for temperatures between
5 and 15°C (Krüner and Rosenthal, 1983; Miller and Libey, 1985; Rogers and
Klemetson, 1985; Bovendeur and Klapwijk, 1989). Experiments with municipal
water report between 0.5 and 2 g TAN m − 2 day − 1 (Anderson et al., 1994; Boller
et al., 1994; Thorn et al., 1996; Parker et al., 1997). Nitrification rates of 1.8 mg
TAN m − 2 day − 1 are reported for highly effective trickling filters at temperatures
of 13–15°C (Anderson et al., 1994). Rusten and Lorentsen (1992), using municipal
wastewater, measured a nitrification rate of 1.6 g TAN m − 2 day − 1, in an upstream
moving bed bioreactor filled with Kaldnes rings, with temperatures varying from 8
to 16°C. Our nitrification rate values lie between 0.1 and 0.2 g TAN m − 2 day − 1
and correspond partly with the earlier experiments in trickling filters at the same
temperatures. The nitrification rates measured in aquaculture lie under the values
found in municipal wastewater, and this is probably a result of the low TAN
concentrations inhibiting nitrification. Also, trickling filters normally show lower
nitrification rates than other biofilters (Speece, 1973; Wortman and Wheaton,
1991).
The lowest nitrification rate in our experiments was obtained in the columns with
the largest specific surface area (Leca). This was not expected, since a larger surface
will lead to a larger area of biofilm. The reason for the low nitrification rate is that
there was 100% nitrification in all the columns. Therefore, in the lower parts of the
columns there was no, or very low, TAN substrate.

5. Conclusion

One hundred percent nitrification was obtained in the columns filled with Leca.
Kaldnes rings showed :80%, and Finturf artificial grass showed the lowest with
36%.
Maximum hydraulic load was only reached for the smallest Leca fractions (2–4
mm) with 0.5 1 min − 1 or 91 m3 m − 2 top surface area of the trickling column per
day.
 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199 197

The nitrification rate measured was between 0.1 and 0.2 g TAN m − 2 day − 1. It
was highest for the Norton rings and lowest for Leca. This value, however, must
be viewed with the knowledge that all the TAN was removed in the Leca filters,
and the removal rate is dependent on the amount of TAN.
The nitrification stabilized 7 – 8 weeks after start-up in the Leca and artificial
grass columns. For the two other media no positive stabilization was observed
over the duration of the experiment (140 days).

Acknowledgements

This study was made possible with grants from the Norwegian Research Council.
The authors would like to thank H. Herland and H. Stokken for technical
assistance and Dr J. Greatorex for critical reading of the manuscript.

References

Anderson, B., Aspegren, H., Parker, D., Lutz, M., 1994. High rate nitrifying trickling filters. Water Sci.
Technol. 9, 47–52.
Ask, M., 1996. Testing of Different Media in Trickling Filters. University of Agriculture, Norway
Master’s thesis, In Norwegian.
Boller, M., Gujer, W., Tschui, M., 1994. Parameters affecting nitrifying biofilm reactors. Water Sci.
Technol. 29, 1–11.
Bovendeur, J., Klapwijk, A., 1989. Fixed-film reactors in water recirculating aquaculture systems:
nitrification kinetics, design criteria, and expected performance. In: De Pauw, Jaspens, Ackefors, H.,
Wilkins, J. (Eds.), Aquaculture — A Biotechnology in Progress. EAS, Bredene, Belgium.
Bovendeur, J., Zwaga, B.A., Lobee, B.G.J., Blom, J.H., 1990. Fixed-biofilm reactors in aquacultural
water recycle system, effect of organic matter elimination on nitrification kinetics. Water Res. 24,
207 – 213.
Bovendeur, J., 1989. Fixed-Biofilm Reactors Applied to Wastewater Treatment and Aquacultural Water
Recirculating System. Agricultural University, Wageningen Ph.D. thesis.
Brunauer, S., Emmet, P.H., Teller, E., 1938. Adsorption of gasses in multi-molecular layers. J. Am.
Chem. Sci. 60, 309–319.
FAKTA, 1995. Values from municipal waste water plants in Norway. Norwegian Pollution Control
Authority Information 11. in Norwegian.
Ferchini, M., Ghrabi, A., Grasmick, A., 1994. Urban waste water treatment by trickling filter and
rotating biological reactor. Water Res. 28, 443 – 473.
Grady, C.P.L., Lim, H.C., 1980. Biological Wastewater Treatment Theory and Applications. Marcel
Dekker, New York, p. 963.
Hem, L.J., Rusten, B., Ødegaard, H., 1994. Nitrification in a moving bed biofilm reactor. Water Res. 28,
1425 – 1433.
Hirayama, K., 1974. Water control by filtration in closed culture systems. Aquaculture 4, 369 – 385.
Kaiser, G.E., Wheaton, F.W., 1983. Nitrification filters for aquatic culture systems: state of the art. J.
World Maricult. Soc. 14, 302–324.
Klemetson, S.L., Cohen, D., 1984. System analysis and design of biofilters for the three components of
macrobarschium production. U.S. Israel Binational Agricultural Research and Development Fund,
Washington DC Project Number US-60-80 of BARD.
KMT, 1996. Information From the Manufacturer Kaldnes Environmental Technology, Tønsberg,
Norway.
198 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199

Knoph, M.B., 1995. Toxicity of ammonia to Atlantic salmon (Salmo salar L.). Department of Fisheries
and Marine Biology, University of Bergen Ph.D. thesis.
Krüner, G., Rosenthal, H., 1983. Efficiency of nitrification in trickling filters using different substrates.
Aquacult. Eng. 2, 49–67.
Liao, P.B., Mayo, R.D., 1974. Intensified fish culture combining water reconditioning with pollution
abatement. Aquaculture 3, 61–85.
Malone, R.F., Manthe, D.P., 1985. Chemical addition for accelerated nitrification of biological filters in
closed blue crab shedding systems. In: National Symposium on the Soft-Shelled Blue Crab Fishers,
Ocean Springs, MS.
Metcalf, L., Eddy, H.P., 1991. Wastewater Engineering: Treatment, Disposal and Reuse, 3rd ed.
McGraw-Hill, New York.
Miller, G.E., Libey, G.S., 1985. Evaluation of three biological filters suitable for aquacultural applica-
tions. J. World Maricult. Soc. 16, 158–168.
Muir, J.F., Roberts, J.R., 1982. Recent Advances in Aquaculture. Westview, Boulder, CO.
Nijhof, M., 1995. Bacterial stratification and hydraulic loading effects in a plug-flow model for nitrifying
trickling filters applied in recirculating fish culture systems. Aquaculture 134, 49 – 64.
Nordeidet, B., Rusten, B., Ødegaard, H., 1993. Phosphorus requirement for tertiary nitrification in a
biofilm. In: Second International Specialised Conference on Biofilm Reactors, EWPCA-IAWQ,
Paris. Poster presentation.
Parker, D.S., Jacobs, T., Bower, E., Stowe, D.W., Farmer, G., 1997. Maximising trickling filter
nitrification rates through biofilm control: research review and full-scale application. Water Sci.
Technol. 36, 255–262.
Roberts, J., 1985. Mathematical models for trickling filter process. In: Jorgensen, Gromiec, G.L. (Eds.),
Mathematical Models in Biological Wastewater Treatment. Elsevier, Amsterdam.
Rogers, G.L., Klemetson, S.L., 1985. Ammonia removal in selected aquaculture water reuse biofilters.
Aquacult. Eng. 4, 135–154.
Rosenthal, H., 1993. The history of recycling technology. A lesson learned from past experience? In:
Reinertsen, H., Dahle, L.A., Jørgensen, L., Tvinnereim, K. (Eds.), Fish Farming Technology.
Balkema, Rotterdam, pp. 341–349.
Russo, R.C., Smith, C.E., Thurston, R.V., 1974. Acute toxicity to rainbow trout (Salmo gairdner). J.
Fish. Board. Can. 31, 1653–1655.
Rusten, B., Lorentsen, A., 1992. Pilot Experiment with Nitrogen Removal at Nordre Follo Municipal
Plant, Post Denitrification with Submerged Biofilter. SFT, Oslo FAN-R 5/92, in Norwegian.
Rusten, B., Hem, L.J., Odegaard, H., 1995. Nitrification of municipal waste water in moving-bed biofilm
reactors. Water Env. Res. 67, 75–86.
SECL, 1983. Summary of Water Quality Criteria for Salmonid Hatcheries, revised ed. Department of
Fisheries and Oceans, Vancouver SECL 8067.
Selseth, 1996. Information from Supplier Rolf Selseth Aquaculture, Stjørdal, Norway.
Skjølstrup, J., Nielsen, P.H., Frier, J.O., Mclean, E., 1997. Biofilters in recirculating aquaculture systems
— state of the art. Scand. Assoc. Agric. Sci. Semin. 258, 33 – 42.
Speece, R.E., 1973. Trout metabolism characteristics and rational design of nitrification facilities for
water reuse in hatcheries. Trans. Am. Fish Soc. 2, 323 – 333.
Sterner, 1997. Information from Supplier Sterner Aquatech.
Thorn, M., Mattsson, A., Sorensson, F., 1996. Biofilm development in a nitrifying trickling filter. Water
Sci. Technol. 34, 83–89.
Timmons, M.B., Losordo, T.M., 1997. Aquaculture Water Reuse Systems: Engineering Design and
Management, 2nd edition. Elsevier, Amsterdam.
van Loosdrecht, M.C.M., Tijhuis, L., Wijdieks, A.M.S., Heijnen, J.J., 1995. Population distribution in
aerobic biofilms on small suspended particles. Water Sci. Technol. 31, 163 – 171.
Van Rijn, J., Rivera, G., 1990. Aerobic and anaerobic in an aquaculture unit — nitrite accumulation as
a result of nitrification and denitrification. Aquacult. Eng. 9, 217 – 234.
Van Rijn, J., 1996. The potential for integrated biological treatment systems in recirculating fish culture
— A review. Aquaculture 3–4, 181–210.
 O.-I. Lekang, H. Kleppe / Aquacultural Engineering 21 (2000) 181–199 199

Vayenas, D.V., Pavlou, S., Lyberatos, G., 1997. Development of a dynamic model describing nitrifica-
tion and nitrification in trickling filters. Water Res. 5, 1135 – 1146.
Wien, A., 1994a. Testing of different biofilters for nitrification in BIOFOR, Part 1: Round media, R&D
report 2-21, VEAS establishment 1991–96. in Norwegian.
Wien, A., 1994b. Testing of different biofilters for nitrification in BIOFOR, Part 2: Media with different
degree of crushing, R&D report 2-22, VEAS establishment 1991 – 96. in Norwegian.
Wien, A., 1995. Testing of different biofilters for nitrification in BIOFOR, part 3: 100% crushed: R&D
report 2-23, VEAS establishment 1991–96. in Norwegian.
Wortman, B., Wheaton, F., 1991. Temperature effects on biodrum nitrification. Aquacult. Eng. 10,
183 – 205.