Journal of Hazardous Materials B90 (2002) 311–321

Brewery wastewater treatment in a

fluidised bed bioreactor
Aoyi Ochieng a,∗ , Tom Ogada b , William Sisenda c , Paul Wambua d
a Chemical Engineering Department, University of Cape Town, Rondebosch 7701, Cape Town, South Africa
b Production Engineering Department, Moi University, Box 3900, Eldoret, Kenya
c Chemical and Process Engineering Department, Moi University, Box 3900, Eldoret, Kenya
d Textile Engineering Department, Moi University, Box 3900, Eldoret, Kenya

Received 29 June 2001; received in revised form 9 November 2001; accepted 10 November 2001

Abstract
A hydrodynamic characteristic performance of a three phase fluidised bed bioreactor has been
studied with brewery wastewater. The influence of operating parameters, such as phase hold up,
phase mixing, aspect ratio and superficial gas velocity, on an aerobic biodegradation in a bioreactor
of 0.16 m i.d. and 2.7 m in height, was analysed. A low-density (960 kg/m3 ) support particle with an
internal interstice was employed. The particle and liquid loading were varied in order to determine
the effect of phase hold up on bed homogeneity. The ranges in which particle loading and bed height
affect fluidisation, and consequently chemical oxygen demand (COD) reduction, were determined.
The distributor used in this work was designed such that fluid flow pattern similar to that of a draft
tube was induced in the reactor. The low-density particles enabled cost effective operation at a
relatively low gas superficial velocity (2.5 cm/s). Aspect ratio significantly influenced the overall
bed homogeneity, and the optimum aspect ratio was 10, with volume of the support particles being
21% of the reactor volume. © 2002 Elsevier Science B.V. All rights reserved.

Keywords: Brewery; Bioreactor; Aspect ratio; Hydrodynamics; Phase hold up; Biodegradation.

1. Introduction

Three phase fluidised beds have gained a considerable application in chemical, wastewater
treatment and in biochemical industries. Some of the reasons for their extensive use are
simplicity in construction, low maintenance due to lack of moving parts, high effective
interfacial areas and therefore, a high heat and mass transfer per unit volume [1].

∗Corresponding author. Tel.: +27-216-5040-35; fax: +27-216-8975-79.

E-mail address: [email protected] (A. Ochieng).

0304-3894/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 3 0 4 - 3 8 9 4 ( 0 1 ) 0 0 3 7 3 - 9
312 A. Ochieng et al. / Journal of Hazardous Materials B90 (2002) 311–321

Nomenclature
C constant in Eq. (3)
Hx height of fluidised bed (m)
H/Dc ratio of bed height to column diameter
UL superficial liquid velocity (cm/s)
Ug superficial gas velocity (cm/s)
Umf minimum fluidisation velocity (cm/s)
VL liquid volume (m3 )
Vr reactor volume (m3 )
Vs support particle volume (m3 )

Greek letters
εg gas hold up
εL Liquid hold up
εS solid hold up
ρL density of the liquid (kg/m3 )
ρp density of the support particles (kg/m3 )

Subscripts
L liquid
g gas
mf minimum fluidisation
p particle
r reactor
s solid

Superscript
n constant in Eq. (3)

In particular, there is an increasing use of three phase fluidised bed bioreactors and
bubble columns in biological treatment of wastewater. This is as a result of a need for a
more specialised method of treatment for industrial wastewaters, which occasionally contain
volatile toxic components. Treatment of these toxic substances in open lagoons could be
a health hazard, therefore, the fluidised bed bioreactor becomes a system of choice. The
geometry of the reactor is dictated by the characteristics of the waste to be treated. Success of
the design and the operation of a three phase fluidised bed bioreactor (TFBB) depends a great
deal on the accurate prediction of the fundamental properties of the system, especially, the
hydrodynamics in the reactor as well as the physical and chemical properties of the feed [2].
Although, TFBB is simple to construct and easy to operate, its scale up poses a significant
challenge due to the complex interaction among many parameters. The complexity of the
interdependence of the hydrodynamic aspects and the intrinsic kinetic ones has resulted in
the development of different mathematical models. However, none of the models explicitly
 A. Ochieng et al. / Journal of Hazardous Materials B90 (2002) 311–321 313

describes the bed characteristics to enable accurate scale up [3]. Some attempts have been
made to identify and quantify the design parameters, however, wide scale application of a
fluidised bed bioreactor still poses a challenge due to the fact that the parameters are waste
specific [4].
One of the principal requirements for a bioreactor is the ability of the micro-organisms
to grow on the support particle and remain embedded for significant biodegradation to
take place. The microbial growth rate is so crucial for bioreactor performance that a lot of
attention has been given to this, owing to the fact that bed turbulence and the nature of support
particle appreciably affect microbial attachment. Considering the nature of the waste, it has
been found necessary to adjust nutrient content and pH to enable initial microbial growth
[5,6]. Gas flow rate affects phase mixing and gas hold ups, which are important parameters
that influence oxygen mass transfer [7]. The survival as well as the activity of the aerobic
micro-organisms depends on the amount of oxygen dissolved. A well mixed medium will
give rise to a homogeneous bed that promotes interfacial mass transfer, and this will make
oxygen available to the micro-organisms.
A correlation of gas velocity, phase hold ups and pressure drop can reveal the influence
of the hydrodynamic characteristics on biodegradation. Pressure drop across the bed is
an economic aspect, which must be considered since it determines the amount of power
consumption for fluidisation. Pressure drop per unit height and phase hold ups can be
calculated as follows [2]
dp
− = (εs ρs + εL ρL + εg ρg )g (1)
dz
where εg , εL , and εs are gas, liquid and solid hold ups, respectively; ρ L , ρ g , and ρ s similarly
represent densities; g is acceleration due to gravity.
εs + εg + εL = 1 (2)
Eq. (1) shows that the use of dense particles leads to an increase in pressure drop, which
consequently causes an increase in power consumption. Low density particles reduce this
tendency, however, they are prone to elutriation. A careful design of a gas distributor and a
bioreactor internals, like draft tube, can minimise this problem.
Gas flow rate contributes significantly to the cost of operating a bioreactor and there is a
correlation between gas hold up and the flow rate, given by
εg = CUng (3)
where the value of the constant n varies slightly from unity, depending on the system
employed and flow regime of the operation. A lot of work has been done on hydrodynamics
of two phase and three phase fluidised bed, however, the data concerning the effects of
the hydrodynamic factors on biodegradation is still limited. There is, therefore, a need to
correlate the hydrodynamic operating parameters with the reactor design parameters in
order to achieve a more efficient biodegradation.
This study focuses on the influence of the hydrodynamic characteristics on aerobic
biodegradation of a brewery wastewater. The aim of this work was to study the influence of
the reactor design and hydrodynamic parameters on the chemical oxygen demand (COD)
and biochemical oxygen demand (BOD) reduction.
314 A. Ochieng et al. / Journal of Hazardous Materials B90 (2002) 311–321

Fig. 1. Schematic diagram of the apparatus.

2. The equipment set-up

The major features of the rig are bioreactor column (5.5 m high and 0.16 m i.d.), dis-
engagement cylinder, and settling tank. A flow sheet for the rig is shown in Fig. 1, where
compressed air was injected into the bioreactor through a U-turn pipe, which was necessary
to prevent back flow of liquid into the air rotameter, in case of power failure. A non-return
valve (N) was used, however, it was observed that some types of wastewater react with in-
ternal components of the valve rendering it ineffective [8]. In such an event, the U-turn pipe
would prevent liquid back flow into air rotameter and subsequently into the cylinder of the
compressor. The disengagement cylinder (2) acted as a foam breaker, while any entrained
bubbles would escape from the settling tank (1) without getting into the pump. The tank
with a capacity of 80 l also provided the necessary head for the pump. The bioreactor (3)
and the settling tank were made of Duran glass, while the tank and the cylinder were made
of stainless steel. Sieves were put in place to prevent particles from getting into tank (1).
The design of the distributor plate was such that, there were more orifices at the centre
of the plate. This was necessary to improve phase mixing with the low-density support
particles by simulating a draft tube fluid flow pattern required for low density particles.

3. Experiments

Initial experiments were carried out with tap water and air at ambient conditions in order
to analyse the fluid hydrodynamic aspects like phase mixing, aspect ratio, phase hold up
and minimum fluidisation velocity (Umf ) with an 11 mm diameter support particle. Given
that the particles used were large, phase mixing was studied by visual observation, such a
method of observation has been reported in literature [2]. The minimum fluidisation velocity
 A. Ochieng et al. / Journal of Hazardous Materials B90 (2002) 311–321 315

was taken as the minimum gas velocity that effected uniform phase mixing. Aspect ratio
was altered by changing bed height, at constant column diameter. The volume of feed in
the bioreactor was dependent on the required operation bed height. The loading was such
that the real volume of the particle was in the range of 4–30% of that of the liquid mixture.
Biodegradation experiments were carried out with a brewery wastewater, which was
collected from a brewery industry. Before commencement of the full operation, a fresh
brewery wastewater feed was introduced into the reactor through which air was passed for
a period of 50 days. At this point stability was attained after a COD reduction from 16,000
to 140 ppm. Beyond this period, there was a gradual decrease in microbial concentration,
and very little change in COD. After attaining a well mixed and adapted culture, further
experiments were carried out batchwise with a hydraulic retention time of 4 h. Microbial
concentration was measured by the method used by Lenas et al. [9], whilst the determination
of COD was done by the standard methods of water analysis [10]. Verification of the accuracy
of the measurements was done with potassium hydrogen phthalate for which the theoretical
COD value is known. The error in COD determination was found to be about 7%. Dry mass
concentration, which was determined by measuring the optimal density of the sample with
a spectrophotometer [9], was found to be about 165 g/m3 .

4. Results and discussion

Preliminary experiments with tap water revealed that the movement of the particles was
influenced by the nature of the fluid flow pattern. Quite significantly, the fluid hydrodynamics
was affected by the nature of the design of the gas distributor. Fluid flow pattern was studied
by comparing the performance of two distributors: one with uniform orifice distribution and
another one with many orifices at the centre. The latter, which was later adopted, exhibited
flow characteristics similar to a draft tube, in which the part of the fluid at the centre of the
reactor moves upwards, and the movement was downwards in the annular. Limited radial
mixing was observed and the entire flow characteristic was akin to that of a continuous
stirred tank reactor. The degree of bed turbulence was governed by the gas flow rate and the
particle hold up, and to a remarkable extent, the minimum fluidisation velocity depended
on these two parameters.

4.1. Phase hold ups and bed height

Fig. 2 shows a sudden increase in Umf , when the ratio of the volume of the biomass-free
solid support to that of the liquid mixture (Vs /VL ) was greater than 20%. At higher support
particle loadings (V s /V L > 20%), the particles were at the top part of the liquid. As the
gas velocity increased, most particles got stuck at the top part of the bed, and the value of
the minimum fluidisation velocity tended to infinity.
The particles used here being less dense than water, initially floated on the surface and the
liquid circulation alone could not fluidise the bed. Fluidisation was, therefore, effected by
the gas phase. It has been reported that in such an operation, with porous plate distributor, the
resultant density of gas–liquid mixture is lower than the density of the liquid phase alone.
Thus, this resultant low density enables particles to sink, hence fluidisation starts [11].
316 A. Ochieng et al. / Journal of Hazardous Materials B90 (2002) 311–321

Fig. 2. Variation of Umf with Vs /VL .

However, in this work, it was apparent that fluidisation occurred not due to the reduced
apparent density but rather due to the flow pattern in the column.
There was no indication that the fluidisation occurred due to reduced density because the
bubbles were relatively big, with the diameter almost equal to that of the particles. Further,
since the bubbles were moving downwards, it was evident they were doing so due to the
induced liquid current flow pattern and not due to reduced density of the mixture. Best fit
curves were drawn though the data point and Fig. 3 shows that there was an increase in gas
hold up (εg ) with an increase in static bed height and increase in superficial gas velocity (Ug ).

Fig. 3. Variation of gas hold up with superficial gas velocity.

 A. Ochieng et al. / Journal of Hazardous Materials B90 (2002) 311–321 317

Fig. 4. Variation of aspect ratio with Umf .

It appeared that there was almost a direct proportionality between εg and Ug , and the
tendency was more pronounced at a relatively low Ug (e.g U g < 3.5 cm/s). This observation
is consistent with the results obtained by Wen and Fan [12], in which it was observed that
the relationship between gas flow rate and gas hold up is linear at low gas velocity but tends
to level off at high gas flow rates. There was a change from homogeneous to churn-turbulent
regime and plug flow characteristics were exhibited at high Ug .
The concept of bed height, per se, is not sufficient to precisely characterise the per-
formance of a bioreactor. Bed height to column diameter ratio (aspect ratio) influences
the fluid hydrodynamics in such a way that when the ratio is high, bubbles have to travel
long distance up the column and as a result, there is more increased chances of coales-
cence. Bigger bubbles move faster resulting in short residence time and consequently low
gas hold up. This fact can explain the decrease in gas hold up with the increase in bed
height as observed in Fig. 3. The decrease in bubble size with the increase in column
height is due to the decrease in pressure with height. Li et al. [13] used a similar reactor
with a perforated gas distributor, and found that bubble size decreases with an increase in
pressure.
From Fig. 4, it can be observed that there was a minimum value of Umf at which the
value of Hx/Dc (Hx = 1.9 m; D c = 0.2 m) was about 9.5. Uniform phase mixing took
place at this point. The left hand side of the graph represents a situation in which liquid
was added to a constant mass (3 kg) of support particles. At the initial stage, the reactor
behaved like a fixed bed, and the value of Umf was very high. This situation persisted as
long as the liquid hold up was still low. As liquid hold up increased, a point (valley) was
reached when the particles could be suspended at a very low gas flow rate, and this was the
minimum Umf (about 0.3 cm/s). The right hand side was due to the increase in bed height
owing to the increasing liquid loading, this resulted in high pressure drop. Higher gas flow
rate was therefore, required to fluidise the bed.
Joshi et al. [14] gave a correlation for gas hold up which showed that Dc (and not H) affects
the gas hold up, however, the correlation contained the bubble velocity. This, however, does
not contradict the present result, since bubble velocity increases with bed height, it follows
therefore that there is an implicit correlation between bed height and gas hold up.
318 A. Ochieng et al. / Journal of Hazardous Materials B90 (2002) 311–321

Fig. 5. Variation of COD reduction aspect ratio.

The decrease in biodegradation with increase in aspect ratio as shown in Fig. 5 confirms
the fact that there was poor phase mixing at higher aspect ratios. Fig. 6 shows that there
was an initial increase in biodegradation with increasing Vs /VL . This increase could be due
to the fact that, as more particles were added, there was an increase in surface area for
micro-organism attachment.
However, as the particle loading increased, there reached a point when the solids in-
hibited phase mixing, and this led to a decrease in mass transfer. Under such conditions,
aerobic micro-organisms are likely to die due to oxygen deficiency leading to decrease in
biodegradation as depicted by the last part of the curve. Optimum Vs /VL was about 21%,
and this result compares quite well with result obtained in Fig. 2, and is consistent with the

Fig. 6. Variation of COD reduction with Vs /VL .

 A. Ochieng et al. / Journal of Hazardous Materials B90 (2002) 311–321 319

result obtained by [8]. With 9 vol.% of support particles, it was observed that the optimum
substrate (liquid waste) loading required was at 2.5 m bed height. This was a trade off be-
tween the rate of biodegradation and the substrate loading. The pressure drop across the bed
does increase with increase in bed height. Increase in bed pressure drop results in decrease
in the stirring effect, it is therefore expected that there should be a decrease in phase mixing
with an increased bed height.

4.2. Superficial gas velocity and gas hold up

Fig. 7 shows that there is a coincidence in the maximum percentage biodegradation for
gas velocity and gas hold up. The optimum gas flow rate was about 2.7 cm/s and the ratio
of Umf /Ug was about 0.6. COD reduction was very low at U g > 4 cm/s, this observation is
consistent with the results of Chen et al. [15], where the column used had an aspect ratio
of 11.8, with particles of density 1250 kg/m3 , which are comparable to the present work.
Similarity of the response of biodegradation to Ug and εg confirms the result in Fig. 3, which
indicates that there is almost a linear relation between gas hold up and gas velocity at low gas
flow rate. It was unlikely that further increase in the gas hold up could cause the decrease in
biodegradation as might be portrayed in Fig. 7, rather, the decrease in biodegradation was due
to the increase in the gas velocity, which resulted in bubble coalescence. As expected, bubble
coalescence reduces mass transfer and this is what caused the decrease in biodegradation
at higher gas flow rates.

4.3. Superficial liquid velocity

Apparently, superficial liquid velocity did not significantly affect COD reduction, how-
ever, there was some increase in COD reduction with increase in UL as shown in Fig. 8.
This tendency could be attributed to an increase in mass transfer caused by increased liquid
turbulence. However, the effect on biodegradation was not significant due to the fact that
the liquid turbulence caused by high fluid flow rate was dampened at high bed height. It

Fig. 7. Effect of gas velocity and gas hold up on biodegradation.

320 A. Ochieng et al. / Journal of Hazardous Materials B90 (2002) 311–321

Fig. 8. Variation of liquid velocity with COD reduction.

was also observed that turbulence caused by liquid flow was not very significant beyond
1.5 m above the liquid injection point.

5. Conclusions

High aspect ratio results in low biodegradation, and the optimum bioreactor geometry
was such that the aspect ratio was about 10 and Vs /VL was 20%. The results indicate that
under similar conditions as of the present work, it would be more economical to operate a
biodegradation process in a wide column diameter with the aspect ratio not more than 10.
The optimum superficial gas and liquid velocities were 2.7 and 0.4 cm/s, respectively. The
results revealed that biodegradation increased with particle loading up to a maximum and
decreased thereafter.
The result obtained showed that the high cost of operating a fluidised bed bioreactor, as a
result of pumping liquid and gas, can be reduced by using low-density particles. However, if
the low-density support particles are used, the aspect ratio should be low in order to achieve
bed homogeneity at low gas flow rates. These results can be useful for a bioreactor design
and the operation.

Acknowledgements

The work was supported by the German Academic Exchange Programme (DAAD).

References

[1] J. Sotelo, F. Beritez, J. Betran-Heredia, J. Rodrueguez, Gas hold up and mass transfer coefficients in bubble
columns. 1. Porous plate diffusers, Int. Chem. Eng. 34 (1) (1994) 82–89.
[2] L. Fan, Gas–Liquid–Solid Fluidisation Engineering, Batterworth Publishers, New York, 1989, p. 35.
[3] T. Mazzarimo, Chem. Eng. J. 36 (3) (1989) 151–159.
 A. Ochieng et al. / Journal of Hazardous Materials B90 (2002) 311–321 321

[4] G.M. Reinhold, F. Lenneman, H. Mark, Modelling and liquid mixing behaviour of biogas tower reactor,
Chem. Eng. Sci. 51 (17) (1996) 4065–4073.
[5] P. Vaananen, T. Kulmala, P. Pouttu, Joint treatment of industrial and municipal wastewater-case-project City
of Kotka, Finland, Water Sci. Technol. 25 (1) (1992) 83–92.
[6] W. Verstraete, Aerobic activated sludge, Biotechnology 8 (1986) 43–76.
[7] S.A. Shetty, M.V. Kantak, Gas-phase backmixing in bubble column reactors, AIChE J. 38 (7) (1992) 1013–
1026.
[8] A. Ochieng, M. Njenga, M. Halfani, Biodegradation of brewery and petroleum wastewaters, in: Proceedings
of the Mechanical Engineering Seminar, Jomo Kenyattaa University of Agriculture and Technology, Kenya,
1998.
[9] P. Lenas, B.C. Baltriy, G.A. Lewandowski, Y.-F. Ko, Biodegradation of the waste in cyclically operated
reactor: theory, experimental verification and optimisation studies, Chem. Eng. Sci. 49 (249) (1995) 4547–
4561.
[10] M. Teres, G. Arnold, R. Hoak, C. Rand, Standard Methods for Examination of Water and Wastewater, America
Public Health Association, 1971.
[11] B. Atkinson, P. Cooper, Biological Treatment of Water and Wastewater, Ellis Howoord Ltd., New York, 1981.
[12] T. Wen, L. Fan, Steady-state phenol degradation in draft tube gas-liquid-solid fluidised bed bioreactor, AIChE
35 (3) (1989) 355.
[13] Y. Li, J. Zhang, L.-S. Fan, Discrete-phase simulation of single bubble rise behaviour at elevated pressure in
a bubble column, Chem. Eng. Sci. 55 (2000) 4597–4609.
[14] J.B. Joshi, V.G. Pargaker, S.D. Gharat, S.V Jadhev, Heat and mass transfer in a multiphase reactors, in: N.
Chereminoff (Ed.), Handbook of Heat and Mass Transfer, Vol. 4, 1990, Gulf.
[15] R.C. Chen, J. Reese, L.S. Fan, Flow structure in a three-dimentional bubble column and three-phase fluidised
bed, AIChE J. 40 (7) (1994) 1093–1101.