Content uploaded by Herman Helness
Author content
All content in this area was uploaded by Herman Helness on Aug 07, 2015
Content may be subject to copyright.
1
High rate biological/chemical treatment based on the
moving bed biofilm process combined with coagulation
H. Ødegaard1, B. Gisvold2, H. Helness2, F. Sjøvold2 and L. Zuliang1
1 Norwegian University of Science and Technology, N-7491 Trondheim, NTNU, Norway
hallvard.odegaard@bygg.ntnu.no
2 SINTEF Civil and Environmental Engineering, N-7465 Trondheim, Norway
Abstract.
A high-rate wastewater treatment process for secondary treatment has been
investigated, consisting of a highly loaded moving bed biofilm reactor directly
followed by a coagulation/floc separation step. It is demonstrated that the biofilm
in such a process mainly deals with the soluble organic matter while coagulation
deals with the particulate/colloidal matter. The bioreactor may, therefore, be
designed based on the soluble COD loading only, resulting in a very compact
plant, especially when a compact separation reactor (i.e. filtration) is used as well.
1. Introduction
Compact wastewater treatment processes are being looked for by cities all over
the world as land available for treatment plants is becoming scarce. In many of the
cities secondary treatment standards are to be met. Direct coagulation/-
flocculation/floc separation, results not only in a very substantial removal of
suspended and colloidal matter, but good removal of organic matter, bacteria and
viruses and micropollutants as well [1,2,3]. These compounds are either associated
with particles and colloids or soluble, high molecular weight organic substances
However, in order to meet secondary treatment standards it may be necessary, to
remove low molecular weight, soluble organic matter as well. For this purpose
biological processes are most suitable from an economic point of view. In the
ambition to make compact treatment plants, biofilm reactors are especially
suitable. However, many of the most compact biofilm reactors (like the granular
media biofilters) cannot accept a high load of particulate matter, since this would
easily clog the filter and result in too frequent filter washing. It is, therefore,
normal to use a two-step process comprised of a pre-coagulation step followed by a
biofilter step. By utilising lamella-separators for floc separation in the primary
step, such treatment plants can be made very compact [4]. Nevertheless one is
reluctant to use a very high organic load because of the fear of clogging of the
biofilters. An alternative will be discussed in this paper, namely to combine the
new moving bed biofilm process with direct coagulation. This process can accept
both a high particulate load as well as a high soluble organic load.
2
2. The moving bed biofilm reactor (MBBR)
In the moving bed biofilm reactor (Figure 1) the biomass grows on carriers that
move freely in the water volume of the reactor, kept within the reactor volume by a
sieve arrangement at the reactor outlet. In aerobic processes, the biofilm carrier
movement is caused by the agitation set up by the air, while in anoxic and
anaerobic processes a mixer keeps the carriers moving (Figure 1a and b). The
biofilm carrier (K1) is made of high density polyethylene (density 0,95 g/cm3) and
shaped as a small cylinder with a cross on the inside of the cylinder and “fins” on
the outside (Figure 1c). The cylinder has a length of 7 mm, and a diameter of 10
mm (not including fins). Lately a larger carrier (K2) of similar shape (length and
diameter about 15 mm) has been introduced as well, intended for use in plants with
coarse inlet sieves and especially for upgrading of activated sludge plants.
K1
K2
a. Aerobic reactor b. Anoxic and anaerobic reactor c. The biofilm carriers
Fig. 1. The moving bed biofilm reactor principle and shape of biofilm carriers
One of the important advantages of the moving bed biofilm reactor is that the
carrier filling-fraction (% of reactor volume occupied with carriers in empty tank)
may be subject to preferences. The standard filling fraction is 67 %, resulting in a
total, specific carrier area of 465 m2/m3 with the K1 carrier. Since the biomass
grows primarily on the inside of the carrier, the effective specific surface area is
335 m2/m3 for the K1 carrier and 210 m2/m3 for the larger K2 carrier at 67 % filling
fraction. It is recommended that filling fractions should be below 70 %. One may,
however, use as much as needed below this, which is convenient, especially when
upgrading plants – for instance from activated sludge to moving bed reactors.
The moving bed biofilm process has been used for many different applications
[5],[6]. In this paper we shall only discuss the high-rate moving bed biofilm
process. It has been shown that the carriers K1 and K2 performs equally well when
comparisons are made based on the effective biofilm surface area [8].
The high-rate moving bed process was also investigated earlier [7], but then an
alternative process scheme based on the use of aluminium-nitrate as coagulant and
anoxic biodegradation was focused on.
3
3. The high-rate Kaldnes moving bed biofilm process
The fate of particles in biofilm reactors is not totally clear, but it is obvious that
particulate matter to a far lesser extent is being degraded in a high-rate biofilm
process than in a standard activated sludge process. When operating at such a high
organic loading that the maximum COD degradation rate prevails, the COD-
removal will primarily be due to consumption of soluble organic matter. The idea
behind the high-rate Kaldnes moving bed biofilm process (Figure 2), is that the
biofilm is supposed to deal with the soluble organic matter, while the coagulant is
supposed to deal with the separation of the particulate matter, including colloids. If
the new larger carrier (K2) is used, pre-treatment can consist of pre-sieving (3-4
mm sieve) only and the flow scheme shown in Figure 2 results then in an
extremely compact plant, especially when a compact separation method (i.e.
filtration) is employed.
Floc
se
p
aration
Coagulant
Fig. 2. Schematic flow diagram for the high-rate process discussed
When used for BOD-removal only, the process has traditionally been designed
for a volumetric loading rate of 4-5 kg BOD7/m3d at 67 % carrier filling fraction
(effective specific surface area: 335 m2/m3) and 15oC. This corresponds to ca 15 g
BOD7/m2d, which is somewhat higher than for other biofilm processes (like RBC)
for the same purpose. In the experiments that are reported here, the performance of
the process at much higher organic loadings was investigated.
4. Biodegradation of soluble organic matter
The biodegradation experiments were primarily carried out in order to
investigate the influence of carrier size and shape in the moving bed process. This
is published elsewhere [8]. Here we shall concentrate on the influence of organic
matter loading on biodegradation and only report the results with the carriers K1
and K2. The experiments were conducted in pilot plants in two parallel lines each,
consisting of one moving bed reactor and one settling tank. The volumes of the
bioreactors were 20 l. The moving bed reactors were operated at organic loads in
the range of 10-120 g COD/m2d and 5-45 g SCOD/m2d. The experiments were
carried out in three different periods. In the first one both reactors were given the
same volumetric load at a filling fraction of 60 %, while in the second period the
filling fraction was varied to give the same effective area load at constant flow.
4
The third period was devoted to a comparison between the two Kaldnes carriers
(K1 and K2) at 70 % filling fraction. The two lines were operated in four sub-
periods at close to constant flow in each period (e.g. the same residence time) and
hence the same volumetric loading rate. The flow of the four periods corresponded
to average residence times of 380, 52, 27 and 18 min. In Table 2 the wastewater
characteristics for the various experimental periods are given.
Table 1. Average, maximum and minimum influent values for the pilot plant
Period 1 Period 2 Period 3
Ave. +
st.dev.
Max Min Ave. +
st.dev.
Max Min Ave. +
st.dev.
Max Min
SS 136+98 505 53 152+55 232 58 88+18 136 53
COD 323+166 893 139 498+235 915 125 219+66 435 119
SCOD 123+39 236 69 219+128 431 36 100+38 211 42
pH 7.3+0.2 8.0 7.0 6.7+0.3 7.1 6.5 7.5+0.1 7.8 7.4
The raw water temperature was in the range of 10-15oC and the oxygen
concentration in the range of 4-6 mg O2/l. Within this range, variations in O2-
concentration are not expected to have any influence on the rate of COD-removal.
4.1. Results from bidegradation experiments
In order to evaluate degradation of organic matter independent of the biomass
separation step, one may look at the removal rate of soluble/filtered COD (SCOD)
versus the loading rate. It is demonstrated in Figure 3 that the maximum removal
rate in this wastewater was found to be around 30 g SCOD/m2d. This maximum
rate was reached at a loading of around 60 g SCOD/m2d. A line through data points
up to this loading is close to linear, indicating that the degradation rate was limited
by the availability of biodegradable organic matter at organic loads lower than this.
0
5
10
15
20
25
30
35
40
0 2040608010
Filtered COD loading rate [g SCOD/m
2
*d]
Filterd COD removal rate
[g SCOD /m
2
*d]
100%
0
0
5
10
15
20
0 2040608
Filte red C OD loadi ng rate [g SCOD /m
2
*d]
Filtered COD removal rate
[g SC OD/m
2
*d]
HRT=380 min
HRT=52
HRT=27
HRT=18
0
a. Period 1 and 2 b. Period 3
Fig. 3. Soluble COD removal rate versus soluble COD loading rate
5
The difference between this line and the 100 % removal line represents,
therefore, the soluble COD that could not be biodegraded in this water within the
actual residence time. The reason for the somewhat poorer removal in the second
part of the experiment (Figure 3b and 4b) stems mainly from the from the fact that
the water was more dilute then with a greater portion of the total COD that was not
biodegradable. Most of the results are also obtained at very low residence times.
We can see, however, from the results of period 3, that there is not much
difference between the results at 18 or 27 min residence time as compared to those
at 52 min. This indicates that the removal of the biodegradable COD is rapid and
that hydrolysis does not play an important part at these short residence times. It is
interesting to note from Figure 3b, however, that at a very low load and long
residence time (380 min - indicated by the line), the slope of the removal/loading
rate relationship is significantly higher than that at the lower residence times (52,
27 and 18 min). This indicates that hydrolysis takes place at this long residence
time.
Above we have concentrated on the removal of soluble (filtered) COD. It is not
easy to analyse the total COD removal rates in the bioreactor, since the
characteristics of both the soluble and the particulate organic matter change
through the reactor by hydrolysis, assimilation etc. In order to be able to take the
particulate matter into account, we have analysed what one may call the
"obtainable" COD removal rate defined as: (CODinfluent-SCODeffluent)*Q/A where Q
is the flow and A is the effective surface area of the carrier. This term illustrates
the removal rate of organic matter if all particles larger than 1 µm was removed in
a downstream separation step. Figure 4 shows that 85-90 % removal of COD could
have been obtained all the way up to a loading rate of 100 g COD/m2d, if the
biomass downstream the bioreactor had been completely removed.
0
25
50
75
100
125
150
0 50 100 150 200
Total COD loading rate [g COD/m
2
*d]
Obtainable removal rate
[g COD/m
2
*d]
HRT=380 min
HRT=52 min
HRT=27 min
HRT=18 min
100%
0
25
50
75
100
125
150
0 50 100 150 200
Total C OD lo adi ng rate [g COD/m
2
*d]
Obtainable removal rate
[g COD/m
2
*d]
100%
a. Period 1 and 2 b. Period 3
Fig. 4. “Obtainable” removal rate versus total COD loading rate
The results demonstrate that a much higher design load than normally used for
secondary treatment may be accepted if efficient biomass separation is assured. In
highly loaded plants clarification of the biomass may, however, represent a
problem. In order to investigate this, another pilot study was carried out.
6
5. Separation of the biofilm by settling
Settleability experiments have been carried out in both jar- and pilot scale on
effluents from high-rate moving bed reactors. In the jar-test experiments an
apparatus based on continuous in-line mixing of the coagulants and pipe
flocculation were used [9]. The pipe flocculators were operated at decreasing G-
values (385 sec-1 for 21 sec and 135 sec-1 for 47 sec) and the flocculated suspension
was introduced to the 2 l jars from the bottom at a flow of 1 l/min. All experiments
were performed with 15 min of slow mixing (25 rpm) and 15 min settling except
for the ferric chloride tests where 60 min of settling was applied.
The pilot plant experiments were carried out in the same pilot plants as the
biodegradation experiments (see above). When a coagulant was used, however,
two flocculation chambers (with decreasing paddle speed and 40 min residence
time) were introduced between the bioreactor and the settling tank. The settling
tanks had a diameter of 0.38 m and a settling depth of about 1 m. Hydraulic
bioreactor retention times in the range of 18 to 380 min were used, resulting in
settling tanks overflow rates in the range of 0.05 to 1 m/h.
5.1. Results of settleability experiments
The effect of coagulation on separation of biofilm from a high rate moving bed
reactor is demonstrated in Figure 6, showing results from the jar-test experiments
0 %
20 %
40 %
60 %
80 %
100 %
012345 67
Dose of polymer [mg/l]
Removal of SS
Zetag 67, Medium c harge
Zetag 75, Medium-high c harge
Zetag 78, High charge
0 %
20 %
40 %
60 %
80 %
100 %
0.0 0.1 0.2 0.3 0.4
Dose of coagulant [mmol Me/l]
Removal of SS
PAX
FeCl3
a. Addition of inorganic metal salts b. Addition of cross-linked (25 %),
(PAX - prepolymerised AlCl3) high MW cationic polymer
Fig. 5. Removal of suspended solids in jar tests of a high-rate MBBR effluent
It is demonstrated in Figure 5a that even a relatively small amount of metal,
(about 0,2 mmol), gave dramatic improvement in settleability as compared to no
coagulant addition. Figure 5b demonstrates that reasonably good SS-removal (85
%) could also be obtained by the use of a relatively low dosage (2 mg/l) of a cross-
linked, medium charged, cationic polymer.
7
In Figure 6 and 7 settleability results from the pilot plants are shown. Figure 6
shows the influence of organic loading on the settleability of the biomass for the
case that no coagulant was added. The SS-removal efficiency is given versus the
total COD loading rate on the bioreactor at different overflow rates on settling
tank. One should be careful in interpreting the actual removal percentages since
these settling tanks were very small, but the general picture with respect to
influence of loading rate can be considered to be correct. The different overflow
rates correspond to three levels of flow and consequently different levels of organic
loading on the bioreactor. This makes the analysis a little complicated since both
loading rate and surface overflow rate varied at the same time.
0 %
20 %
40 %
60 %
80 %
100 %
01020
Bioreactor loading [g COD/m
2
*d]
SS-removal in settling tank
v=0.05 m/h
v=0.16 m/h
v=0.16 m/h w/polymer
v=0.35 m/h
0 %
20 %
40 %
60 %
80 %
100 %
02040
Bioreactor loading [g COD/m2*d]
SS-removal in settling tank
v= 0. 05 m / h
v= 0. 35 m / h
v= 0. 65 m / h
60
30
Fig. 6. Influence of organic loading rate Fig. 7. Influence of polymer addition
in bioreactor on settleability on settleability
It is obvious, however, that not only the overflow rate but also the organic
loading on the bioreactor has a pronounced effect on settleability. At a given
overflow rate (for instance 0,35 m/h in Figure 6), there is a decrease in settleability
with increasing organic load. This was the case both when total COD and soluble
COD was considered, indicating that the organic loading regime experienced by
the microorganisms, does influence on the settleability of the biofilm that is
sloughed off the carriers. The consequence of this, from a practical point of view,
is that settling ought to be enhanced by coagulation when one is operating this flow
scheme at high organic loading rates. In Figure 7, results from experiments where
two pilot plants have been run in parallel, are compared in order to study the effect
of adding a cationic polymer coagulant at a given overflow rate. 1,5-2 mg/l of the
medium charged, high MW polymer was added. The regression lines in Figure 7
are drawn for illustration purposes only.
Even though there is considerable scatter in the data, it is demonstrated by the
results at the same overflow rate that; a) settleability is better with polymer
addition and b) settleability is less influenced by the organic loading on the
bioreactor when a polymer is added. The reason for this is believed to be that the
polymer is able to flocculate the smallest particles that are more abundant the
higher the organic loading is. The draw-back of poorer settleability at higher
organic loading is therefore compensated for by the polymer coagulation.
8
6. Separation of the biofilm by direct filtration
Since the amount of sludge that is to be separated, is quite low in the actual
process scheme, direct filtration may be an alternative to flocculation/settling. If
phosphate removal is required, a metal coagulant (Al or Fe) will be necessary, but
if only SS- and BOD-removal is looked for, a cationic polymer alone, or in
combination with a low dose of metal salt, may be used. This would minimise
sludge production and might make the use of direct filtration possible.
A filter for such an application should, of course, be built with a high sludge
retaining capacity in order to achieve acceptable filter run times. In this project we
have used an up-flow filter with expanded clay aggregates (Filtralite) as filter
medium with a wide range of grain sizes (1.5-4 mm). The filter grain sizes will
arrange itself from coarse to fine in the direction of flow after backwashing. This is
ideal in terms of storage capacity and even distribution of particle deposits
throughout the entire filter bed depth. Because of the low density of the lightweight
expanded clay aggregates, this filter is operated with low filter backwash rates.
The filter of the pilot-plant used in our experiments (Figure 8) had a bed depth
of 1.2 m, a 10-cm gravel layer (5-10 mm) as support layer, and 1.5-4 mm crushed
Filtralite as filter medium. The inner diameter of the filter column was 85 mm and
the total available head 1.8m. The head-loss could be determined by a pressure
sensor at the base of the filter column.
Experiments were performed
without coagulant addition as well as
with the addition of various
coagulants - ferric chloride at a dosage
of 10 mg Fe/l, the low MW, highly
charged polyDADMAC (Magnafloc
368) at a dosage of 1 mg/l and the
high MW, highly charged poly-
acrylamide (Floerger FO4440SH) at 1
mg/l. The organic loading on the
bioreactors varied somewhat
throughout the experiments but was
on average 8.0 kg COD/m3*d (36.5 g
COD/m2*d) and 4.5 kg SCOD/m3*d
(20.5 g SCOD/m2*d). Fig. 8. Filter pilot plant
MBBR
1.5-4 mm
H= 1.5 m
Backwash
water
Backwash
outlet
Effluent
H= 3 m
Pressure sensor
Filter area = 5.67*10
-3
m
2
Filter volume = 6.81*10
-3
m
3
H= 1.8 m
H= 1.2 m
6.2. Results of filterability experiments
Figure 9 shows the effluent SS-concentration versus filtration rate. The lines are
drawn for illustration purposes only. It is demonstrated that there is a close to
linear relationship between the effluent SS-concentration and the filtration rate, no
matter what kind of pre-treatment that has been used. The poorest results were
obtained by the use of iron alone while the two different cationic polymers gave
about the same results, which were almost 10 mg SS/l lower that with iron alone
9
0
5
10
15
20
25
30
35
0510
Filter veloc ity [m/h]
Average suspended solids
in e fflu en t [mg S S/l ]
No coagulant 1 mg M368/l 10 m g Fe/l 1 mg FO4440SH/l
0
10
20
30
40
50
60
0.0 0.5 1.0 1.5
Sludge loading rate [kgSS/m
2
*h]
Filter run time to 1 m head loss [h]
No coagulant 1 mg M368/l 10 mg Fe/l 1 m g FO44 40SH /l
15
Fig. 9 Effluent SS-concentration versus Fig. 10. Filter run time to 1 m head-
filtration rate loss versus sludge loading rate
.
Even at filtration rates as high as 20 m/h, the effluent SS-concentration could be
kept under 20 mg/l while only 5 m/h could be used in order to obtain the same
effluent SS-concentration with iron alone. This must be caused by the fact that iron
alone produce small particles through coagulation in addition to those already
present in the water, that escape the filter more easily than those bound together
through the action of the polymer. Iron alone caused weak flocs leading to filter
break-through as the determining factor for filter run time.
The filter run was terminated either when the maximum allowable head-loss
(set at 1m) was reached or when breakthrough occurred (when maximum
allowable effluent SS concentration, set at 30 mg SS/l, was reached). In all runs
with iron alone as coagulant, break-trough determined the length of the filter run
(max allowable effluent concentration of 30 mg/l was reached before max
allowable head-loss) while maximum head-loss determined the length of filter run
when a polymer was used as coagulant.
In Figure 10 the filter run time versus the sludge loading rate is presented. The
lines are drawn for illustration purposes only. It is demonstrated that the cationic
polymer with the lower molecular weight (Magnafloc 368) gave longer filter runs
at a given sludge loading rate than the one with high molecular weight (FO440SH)
at the same dosage. This is caused by the fact the low MW polymer acts as a pure
coagulant, resulting in relatively small, compact flocs, while the high MW polymer
acts according to the bridging mechanism as well, resulting in larger flocs that can
not penetrate equally far into the filter. The fastest head-loss build-up was
experienced with iron alone as a result of the higher amount of sludge to be
separated as a consequence of metal hydroxide precipitation.
At the addition of only 1 mg/l of the low MW, cationic polymer, about 16 hours
filter run time was obtained at a sludge loading rate of about 0,5 kg SS/m2h. A
sludge loading rate as high as 0,75 kg SS/m2h could be used if a filter run time of
10 hours was acceptable. The latter loading equals a filtration rate of about 5 m/h
10
in a case without presettling (SSin filter~150 mgSS/l) . At filtration rates below 10
m/h, the effluent SS-concentration could be expected to be < 15 mg SS/l.
7. Conclusions
The following conclusions can be drawn from this study:
1. The combination of a high-rate moving bed reactor, possibly without primary
settling, followed directly by a coagulation step will result in an extremely
compact wastewater treatment plant, especially when a high-rate separation
method (i.e. filtration) is also employed.
2. If only SS- and BOD-removal is looked for (secondary treatment), a cationic
polymer alone can be used, minimising sludge production. In this case, a low
molecular weight, highly charged cationic polymer should be the chosen. The
necessary dosage can be expected to be in the range of 1-2 mg/l.
3. If phosphate removal is required as well, a metal coagulant has to be used, but
a low dosage is needed, normally < 0,2 mmol/l.
4. If direct filtration is used for floc separation, a sludge loading rate of about 0,5
kg SS/m2h would result in a filter run time of about 16 hrs with the actual filter
5. At filtration rates between 7 and 10 m/h, the effluent SS-concentration could
be expected to be < 15 mg SS/l and < 10 mg SS/l at filtration rates < 7 m/h.
8. References
1. Ødegaard, H.: "Particle separation in Wastewater treatment." Documentation 7' th
European Water Pollution Control Association Symposium, Munich, May, 1987.
2. Ødegaard, H.: "Coagulation as the first step in wastewater treatment". In : Hahn, H.H.
and Klute, R. (eds): Pretreatment in Chemical Water and Wastewater Treatment.
Springer Verlag, Berlin/Heidelberg, 1988, pp. 249-250.
3. Ødegaard,H.: "Norwegian experiences with chemical treatment of raw wastewater".
Wat. Sci. Tech. Vol. 25, No. 12, 1992, pp. 255-264.
4. Pujol, R., Sagberg, P., Lemmel, H. and Hamon, M.: "The use of reagents in up-flow
submerged biofilters" In : Hahn, H. H. and Klute (eds): Chemical water and wastewater
treatment III. Springer Verlag, Berlin/Heidelberg, 1994, pp 221 - 230.
5. Ødegaard, H., Rusten, B., Westrum, T.: "A new moving bed biofilm reactor -
Applications and results". Wat.Sci.Tech. Vol 29, No 10-11, 1994, pp 157-165.
6. Ødegaard, H., Rusten, B. and Siljudalen, J.: "The development of the moving bed
biofilm process - From idea to commercial product. European Water Management Vol.
2, No. 3., 1999, pp 36-43.
7. Æsøy. A., Ødegaard, H. and Sandberg, R.: "Anoxic degradation of dissolved COD for
enhanced organic matter removal in compact chemical treatment plants". In : Hahn, H.
H., Hoffmann, E. and Ødegaard, H. (eds): Chemical water and wastewater treatment
IV. Springer Verlag, Berlin/Heidelberg, 1996, pp 387 - 398.
8. Ødegaard, H.. Gisvold, B. and Strickland, J.:"The influence of carrier size and shape in
the moving bed biofilm process" Wat. Sci. Tech. Vol. 41, No 4-5, 2000.
11
9. Ødegaard, H., Fettig, J. and Ratnaweera, H.: "Coagulation with prepolymerized metal
salts". In Hahn, H.H. and Klute, R. (eds): Chemical Water and Wastewater Treatment.
Springer Verlag, Berlin/Heidelberg, 1990, pp.189-219.
