Predicting the bed expansion of an anaerobic fluidised-bed bioreactor

e

Pergamon 0273·1223(9S)00420-3

Will. ScL r,ch. Vol. 31,No. 9, pp.181-19I,I99.5. Copyrigbt C 199.5 IAWQ Printed in Oreat Britain. All rigbll relCfYOd. 0273-1223~.5 $9'SO +(){)()

PREDICTING THE BED EXPANSION OF AN ANAEROBIC FLUIDISED-BED BIOREACTOR Tjandra Setiadi Department of Chemical Engineering /TB, Inter University Centre on Biotechnology ITB, JL Ganesa 10, Bandung 40132, Indonesia

ABSTRACT In this study, the bed expansion of an anaerobic f1uidised-bed bioreactor was investigated. The experiments were conducted in a 4.4 em diameter colwnn with 0.152 mm sand as mediwn support of the bioftlm. The temperature was maintained at 30"C and synthetic wastewater was used. The bed expansion data from the prescnt study and literature data were used to evaluate the applicability of the correlations In predicting the bed expansion of the bioreactors. A correlation between bed expansion or voidage (e) and the particle Reynold's (Re) and Gallileo's (Ga) numbers proposed by Setiadi (1989)

e = 1.72 ReO.203 Ga -0.179 was found to predict the bed expansion of anaerobic f1uidised-bed bioreactors with better accuracy than any previously published correlations. KEYWORDS

Anaerobic; bed expansion; bed voidage; bioreactor; correlation; fluidised-bed. INTRODUcnON

The application of anaerobic fluidised-bed bioreactors in treating various wastewaters has gained considerable attention in the last two decades, although most of them have been operated only on laboratory and pilot-plant scales. However. since 1983 full-scale anaerobic fluidised-bed reactors have been started up in the USA and The Netherlands (lza, 1991). With the increase in commercial application. deterministic methods for the design of fluidised-bed bioreactors should be developed. In the design of a fluidised-bed reactor, it is important to know the Stllte of bed expansion, as this influences the bed volume and hence the residence time of the liquid phase. The bed expansion has a close relationship to bed voidage, therefore in the paper these terms are used interchangeably. Studies on bed expansion of the fluidi.o;ed-bed reactor have been made extensively for non-biological particles. However. with the increasing application of the reactor to biological· systems, several workers have investigated the bed expansion of fluidised-bed bioreactors. 181

T.SETJADI

182

A fluidised-bed bioreactor is a novel biological system in which small. fluidised particles are used to immobilise bjoma.~ within the reactor. This results in a high biomass concentration in the reactor, enabling it to (lperate at significantly higher liquid throughputs than stirred tank reactors with the absence of significant loss of biomass. To model the kinetics or the dynamics of these biologically active fluidisedbeds, it is important to be able to establish relationships between the liquid superficial velocity UL and the bed voidage E. A number of workers have investigated the relationship between the liquid velocity and the bed voidage £ for anaerobic/anoxic fluidised-bed bioreactors (Ngian and Martin. 1980; Hermanowicz and GancZll!Czyk. 1983; Timmermans and van Haute, 1984; Mulcahy and Shieh. 1987). The gas produced in the reactor may

influence the bed expansion behaviour of two-phase (liquid-solid) fluidised-bed reactors. However, none of the above workers hali reponed any effect of produced gas on the bioreactor systems. This was in accordance with Davison (1988) who attempted to study the effect of low gas velocity up to 1.2 cmls on the bed expansion of gel beads (particle density, Pp 1.1 g1cm3). It was found that the effect of gas flow rate on the bed voidage e was small. Thus, the bed voidage with a low gas velocity fluidised-bed bioreactor can be estimated from that of two-phase (L-S) fluidised-bed reactors.

=

Table I. The summary of fluidised-bed bioreactor studies f1uidisation correlations

Investigators 1. Ngian and Martin (1980)

Richardson and Zaki (1954)

(1981)

4. Timmermans and van Haute (1984) S. Mulcahy and Shieh (1987)

(1966)

• applicable for estimation of the expansion index n. but the usual expression for VI of spherical particles could not always be applied.

2. Shieh e/ 01. 3.lIermanowicz and Ganczarczyk (1983)

Wen and Yu

• applicable for estimation of n, but proposed a new correlation to calculate C. • applicable for estimation ofn and VI • applicable, but proposed new correlation for ~ and C.

• satisfactory predicting the bed expansion. • gave lower bed voidage

A summary of biological fluidised-bed studies is given in Table 1. All the studies on the fluidisation characteristics of biocoated particles used the established correlations for spherical particles. Among them, the correlation of Richardson and Zllki (19j4) was the most widely used. It can be easily understood. because it hali a very simple form and consequently is easy to modify. A modification of the correlation of Richardson and Zald (19j4) was proposed by Hermanowicz and GanczllrCzyk (1983) and Mulcahy and Shieh (19H7) to predict the bed voidage of the biological denitrification systems. In contrast. the correlations of Wen and Yu (1986) and Richardson and :zaki (1954) were suggested to give an excellent prediction of the

Predicting bed expansion of an anaerobic fluidised-bed bioreactor

183

bed voidage of fluidised-bed bioreactors by Shiehet al. (1981) and Timmermans and van Haute (1984), respectively. From the design point of view, the equation of Wen and Yu (1966) is simpler to use, because the settling velocity of the panicle vt is not required. But the form of this correlation has discouraged researchers from attempting to modify it. No study has used non-spherical particle correlations for predicting bed expansion behaviour. The reason is that to determine the shape factor of biocoated particles is difficult. The simple method of measuring sphericity suggested by Cleasby and Fan (1981) can not be applied to biological particles. Table I also reveals that each correlation is specifically valid for its own data. No attempt has been made to generalise the correlation to cover all sets of data. Additionally, bed expansion data of anaerobic fluidisedbed bioreactors are rather limited. Therefore, there is a need for further data on the bed expansion of anaerobic fluidised-bed bioreactors. Moreover. it is desirable to have a general correlation to predict the bed expansion of the bioreactor. EXPERIMENTAL The experimental set up is shown in Fig. 1. The major components of the system were the reactor, gas-liquid separator, gas coIlector, liquid pump and water bath with a temperature controIler. The reactor was constructed from two concentric Plexiglas columns of different diameter. The inner column had 4.4 cm 10, 200 cm length, and 0.3 cm wall thickness and the outer had 5.7 cm 10, and 0.3 cm wall thickness. Scales graduated to one miIlimetre were fixed to the column. The water from a water bath was circulated in the jacket space between the two columns to maintain the temperature inside the column at around 30°C. At the bottom there was a 23 cmchigh calming section filled with glass beads of 1.5 mm diameter. Five equally spaced sampling points were installed along the column length to obtain both liquid and bed samples. The first sampling point was located 35 cm from the bottom of the column and the others were at distances of 30 cm from one to another. A 1 litre gaslIiquid separator was placed on the suction side of the recycle pump to facilitate the dissolution of gas from the recycled flow. The gas produced in the reactor was collected in a I litre gas collector filled with acidified saturated NaCI solution. so that the volume of the dLl\placed salt solution was equal to the volume of gas produced. The experimental procedure was divided into three parts i.e., inoculum preparation. reactor start-up, and bed expansion experimentation.

Inoculum preparation. Cow dung was collected from a farm and screened through a sieve to remove coarse material. Then I litre of the sludge was transferred to a 2 litre flask containing 100 ml of synthetic wastewater (Chen et al., 1985) and purged with N2' The flask was placed in a temperature controlled room of 30°C. Synthetic wastewater (100 ml) was added to the flask each day to finally obtain 2 litres in the flask. After the 2 litre volume was obtained, 100 ml sludge was removed each day from the flask and replaced with 100 ml synthetic wastewater to maintain a constant volume. This practice wa.c; continued for about 2 months to ensure that the inoculum was fuIly acclimatised to the synthetic wastewater. Start-up of the anaerobic fluidised-bed bioreactor. The column was filled with 1350 g clean sand with an average diameter of 142.3 m, then 1 litre of inoculum was added, and the column was filled with synthetic wastewater. The column was operated in a total recycle, with initial bed expansion maintained at 30%. To maintain the reactor pH in the range of 6.8 to 7.2, sodium bicarbonate was added as necessary. 200 ml of synthetic wastewater was added to the reactor each day to promote and sustain the growth of biofilm on the particles. Bed expansion experimentation. Throughout the experimental period, the thickness of the biofilm on the surface of the sand particles increased as a result of the growth of the mixed bacteria. At a certain biopanicle size, the expanded bed height of the beds H at different superficial liquid velocity UL were measured. During

184

T.SETIADI

each bed expansion experimentation. the solid samples were withdrawn from three different height of sampling poinL The bioparticle diameter db was measured with the methods given elsewhere (Shieh et al., 1981). The terminal settling velocity of particles was measured by dropping individual particles in a waterfilled Plexiglas column of 11.2 cm 10 and the travelling time between two marks (1 m apart) was recorded. At least SO measurements were made for each determination. From the data of db and H, the bed voidage £ corresponding to each superficial velocity UL was calculated by the equation of Ngian and Martin (1980). The biopartic1e density determination was by the indirect method employed by Hermanowicz and Ganczarezyk (1983).

~

point "'"

d::on..

Figure I. EXperimental set-up for bed expansion study.

RESULTS AND DISCUSSION The determination of the wet density of the biofilm nnd of the bed expansion behaviour of the bioreactor nre discussed firsL' In the final section. n comparison of different correlations and available data of bed expansion of biological fluidised·beds is performed. Determinj1tion of wet density of biofilm The wet density of biofilm for different biofilm thicknesses is presented in Table 2. For biofilm thickness in 11 range of 2.05 to 15.21lrn, the wet density varied from 1.0 to 1.17 glcm 3, with a mean value of 1.09 glcm 3• The value was not much different from that of biofilm in 11 denitrification system. Densities of 1.10 glcm 3

Predicting bed expansion or an anaerobic nuidised-bed bioreactor

18~

and 1.14 g1cm 3 have been reported by Ngian and Martin (1980) and Hermanowicz and Ganczarczyk (1983), respectively. Table 2. Wet density of biofilm and density of biocoated particles

Biofilm Thickness (J.Im) 2.05 6.1 11.3 15.2

Biofilm Densit~,

(g1cm ) 1.0 1.08 1.13 1.17

Pwb

Bioparticles diameter (J.Im) >

Density of Biocoated Particles, Pb (glcm3)

146.4 154.5 164.9 172.7

2.52 2.31 2.11 1.98

The density of biocoated particles decreased with increasing biofilm'thickness in that it decreased from 2.52 g/cm3 for 2.05 11m biofilm thickness to 1.98 g/cm 3 for 15.2 I!m biofilm thickness. Thus, an increase of biofilm of just over 10 I!m would lead to a 20% decrease in the bioparticle density. The effect of biofilm thickness on the particle density would influence the bed expansion characteristics of a fluidised-bed bioreactor. This will be discussed in the following section. Bed expansion characteristics of an anaerobic fluidised.bed bioreactor Data on the expanded bed height H versus liquid superficial velocity uL as the biofilm thickness increased are given in Fig. 2. An expansion of the bed as the velocity increased was observed. The effect of liquid velocity was more pronounced on the thicker biofilm, since the density of bioparticles decreased with increasing biofllm thickness (Table 3). That is, at liquid velocity UL = 0.183 crn/s, the bed height was 111.6 cm for 2.05 I!m biofilm thickness compared with 186.3 cm for IS.2I!m biofllm thickness. Therefore, a 13 J.lffi increase in biofilm thickness caused a 67% increase in the bed height. This is an imporUnt result from the point of view of reactor design. A slight change in bioparticles diameter sizes. from 146.4l!m to 172.7 J.lffi or an 18% increase led to an increase in reactor height of more than 60%. Furthermore. the increase of bed height gave rise to an increase in liquid residence time in the reactor by the same magnitude; thus it would affect the reactor performance. The expanded bed height H data could be easily transformed to bed voidage e data by using the following equation (Ngian and Martin, 1980): (1)

where

w: db:

total dry weight of clean sand particle diameter of bioparticles dm: diameter of clean sand H: . bed height A: cross-sectional area of the bed Pp: density of clean sand.

A typical plot of bed voidage versus liquid superficial velocity is presented in Fig. 3. The bed voidage e 0.055 cmls to 0.768 at UL =0.365 cmls for 11.3 11m biofllm thickness. A SimIlar trend was found for different biofllm thicknesses (Fig. 4). Although, there was a great difference in bed height with increasing biofllm thickness. the effect of biofllm thickness on bed voidage was almost insignificant. The difference in bed voidage at the same liquid velocity for different biofilm thicknesses was i~cr;ased from 0.495 at uL

=

T.SETlADl

1116

nlways less than 5% (Fig. 4). This was because the increase in biofilm thickness also caused an increase in the total volume of bioparticles. Thus, the bed voidage, which £

wa.~

•

1·

total vol. ofbiopartic1es

._

_.__ . total expanded bed

(2)

not affected greatly by the increase in total bed height. 200

,.-----,------r----,------,

180

-a -t

160

u

140

t:I

M

= Q

fat P3

120 100

80

BIOFILM THICKNESS

o = 2.05 /lm • = 6.10 /lm 'V = 11.3 p,m y. = 15.2 p,m

60 '--_ _---'l....-_ _--' ..... --' 0.0 0.1 0.2 0.3 0.4 LIQUID SUPERFICIAL VELOCITY (em/II) Figure 2. The relationship between bed height and liquid superficial velocity.

Comparison of bed expansion correlations to present study data Despite the fact that no general correlation is available to predict the bed voidage of biological system, the literature correlations (Richardson and Zaki, 1954; Wen and Yu, 1966; Hennanowicz and Oanczarczyk, 19H3 and Mulcahy and Shieh, 1987) were used to predict the bed voidage data from the present study. Desides the literature correlations, the correlation proposed for non-spherical particles by Setiadi (1989) was also used. The correlation of Setiadi (1989) is a correlation of bed voidage (e) and the particle Reynold's (Re) and Oa11i1eo's (Oa) number:

e =1.72 ReO.203 Oa -0.179 where

Re Oa

p u g

=

= = = III

(3)

UL·dt/u

db(Pb. p).g!(U2.p)

liquid density liquid kinematic viscosity acceleration of gravity

The correlations Were compared to the actually measured bed voidage for particles with 1l.3 ~m biofilm thicknes.". The comparison is presented in Fig. 5, although 5 correlations were tested, only four curves are presented in the figure. The reason wa.'1 that the correlations of Richardson and Za1d (1954) and

187

Predicting bed expansion of an anaerobic fluidised-bed bioreactor

Hermanowicz and Ganczarczyk (1983) predicted the same values of bed voidage. The figure shows that the correlations of Richardson and Zaki (1954), Hennanowicz and Ganczarczyk (1983), and Wen and Yu (1966) predicted a lower value of bed voidage than that obtained experimentally. In contrast, the correlation of Mulcahy and Shieh (1987) predicted a much higher value of bed voidage. The correlation of Setiadi (1989), however, gave good agreement with the experimental data. O.B

r---------r--~---_r___----_..

0.6

BIOFILM THICKNESS

o = 11.3 p.m 0.5

1....-

0.0

....1.-

0.1

--1-

0.2

LIQUID SUPERFICIAL VELOCITY (em/I)

--1

0.3

.Figure 3. The relationship between bed voidage and liquid superficial velocity.

The comparison of different correlations and experimental data was simplified by presenting it in tenns of the sum of squares and mean deviation as shown in Table 3. The values given in the table were for the number of data points of 36. From the table, it can be seen that the correlation of Mulcahy and Shieh (1987) gave the highest sum of squares and mean deviation. The high deviation of the correlations from the experimental data is difficult to understand, since the correlation was derived from biological f1uidised-bed systems. The probable reason was the difference in biofilm thickness. In the present study data, the maximum biofilm thickness waS 15.2 J.Ull, whereas the correlation of Mulcahy and Shieh (1987) was derived from 40 Ilm to 1200 I!m biofilm thickness data. The second highest sum of squares and mean deviation was given by the correlation of Richardson and Zaki (1954) and was followed by Hermanowicz and Ganczarczyk (1983). Both correlations gave a mean deviation just over II %. The lowest sum of squares and mean deviation among the literature correlations was given by the equation of Wen and Yu (1966). The Setiadi's (1989) correlation for non-spherical particles appeared to predict the bed voidage of the anaerobic f1uidised-bed reactor satisfactorily. The mean deviation ofthe correlation was less than 2%. Thus, the correlation of Setiadi (1989) is proposed as an alternative for predicting the bed voidage of biological f1uidised-bed systems.

T.SETIADI

188

Bed expansion correlation foc !l fluidised-bed bio!'CactQC

All the correlations proposed in the literature to predict the bed voidage of fluidised-bed bioreactors were specifically valid for their own data. No attempt has been made to test the applicability of the correlations to different sets of data. Therefore, in this study an attempt was made to compare the available data on bed voidage of fluidised-bed bioreactor with predicted values from different correlations (Richardson and Zaki, 1954; Wen and Yu, 1966; Hermanowicz and Gancznrczyk, 1983; Mulcahy and Shieh, 1987; Setiadi, 1989). Two published sets of data available for biological denitrification systems were used (Ngian and Martin, 19110 and Timmermans and van Haute, 1984).

.----...,-----r-----,.----.....,

0.9

0.8

BIOFILM THICKNESS o 2.05 J.Lrn • = 6.10 J.Lrn t:l = 11.3 J.Lffi ... "" 15,2 IJ.rn

=

0.6 '"

0.5

oi-

L-

0.0

0.1

J.-

0.2

..L-_ _- - I

0.3

LIQUID SUPERFICIAL VELOCITY (em/e)

0.4

Figure 4. Bed voldage as a function of liquid superficial velocity.Comparison of bed expansion correlations to present study data.

Table 3. Sum of squares ond mean deviation of bed voidage for the present study data Correlations Wen and Yu Hermanowicz Mu!cahy& Richardson (1966) & Ganczarczyk (1983) Shieh (1987) and Zaki (1954) I.S.

0.23

0.05

0.22

M.D. (%)

11.S

S.OO

11.4

U.· sum of squares. M.D.· mean deviation (%). t n· number of data points.

1.05 25.7

[I (E ca!· E exp) I/eexp]/n"100 %.

Setiadi (1989) 0.019 1.1

189

Predicting bed expansion of an anaerobic tluidised-bed bioreactor

The available data found in the literature were not readily in the fonn of a bed voidage/liquid velocity relationship. The data of Ngian and Martin (1980) were given in expanded bed height versus liquid superficial velocity. However, this was easily transfonned to bed voidage/liquid velocity fonn using equation 2. The number of data points from this paper was 84 with biofilm thickness up to 70 ~m. 0.9 r-------~-----,.....----___,

(1)

EXPERIMENTAL DATA BIOFILM THICKNESS 11.3 J.Lrn

........

0.8

1'0:1

~ ~

·0.7

..... ..... .......... ,.

....oQ

> Q

f;&1

l:Q

......

0.6 .. "

.....

..

,.

0.5

0.4 '--

0.0

,

,.

..'

.... .-

........... .

0",.••

",

.

•.....".0"

.

," / ../ CURVE 1 MULCAHY &c SHIEH (1967) /./ CURVE 2 SETlADI (1989) ..··'·CURVE 3 RICHARDSON &: ZAKI (1954) . HERMANOWICZ &c GANCZARCZYK (1983) CURVE 4 WEN &c YU (1966) --11--

0.1

--"

0.2

--'

0.3

UQUID SUPERFICIAL VELOCITY (em/a) Figure S, The comparison of experimental data and predicted values of bed voidage

The data of Timmennans and van Haute (1984) had biofUm thickness up to 500 Ilm with a number of data point" of 40. They were presented in terms of biomass concentration. Nevertheless, the biomass concentration is related to the bed voidage, as follows (Shieh et al., 1981):

(4) where

biomass concentration biofilm dry density diameter of support particle diameter of bioparticle.

The third group of data wa." from the present study for an anaerobic fluidised-bed bioreactor (FBR). The number of data points from the present study was 36. The three groups of data, therefore, were used to evaluate the applicability of the correlations. The evaluation was done by calculating mean deviation (%) between predicted value and experimental data and this is presented in Table 4. This shows that the highest mean deviation for total data was from the correlation of Mulcahy and Shieh (1987), followed those of by Richardson and Zaki (1954), Wen and Yu (1966), Hermanowicz and Ganczartzyk (1983) and Setiadi (1989). Each correlation is discussed in this section.

190

T.SETlADI

From Table 4. it can be seen that for the data of Ngian and Martin (1980). the correlation of Mulcahy and Shieh (1987) had the highest mean deviation. A similar finding was also observed for the data of Timmermans and van Haute (1984) and this study. The mean deviation for all of the data was just below 30%. The explanation for the high deviation of the correlation from the experimental data is not clear. The probable reasons are the differences in biofilm thickness and the different techniques in the determination of bed voidage. The biontm thickness. anyhow. was not an issue. since the data of Timmermans and van Haute (1984) had a range of biofilm thickness of 50-500 J.l.m being totally in the range of the correlation of Mulcahy and Shieh (1987). Nevertheless. the mean deviation was still very high: it was just below 50%. The other reason was the different technique in the determination of bed voidage. However. justification of the methods cannot be made so far, because in the present study only the method of Ngian and Martin (1980) wa.~ used to calculate the bed voidage. Therefore. a further study of the different methods of assessing bed voidage is recommended. Table 4 also shows that the spherical particle equation of Richardson and Zaki (1954) gave a mean deviation of 12.1%. 18% and 11.5% for the data of Ngian and Martin (1980), Timmermans and van Haute (1984), and this study. respectively. The mean deviation for the total data was 13.5%. The correlations of Wen and Yu (1966) gave a lower mean deviation than those of Richardson and Zaki (1954) for the data of Ngian and Martin (1960) and Timmermans and van Haute (1984), but a higher mean deviation for the data of this study. For the total data, the mean deviation was 10.1 %. The lowest mean deviation for total data among the published correlations was given by the correlation of Hermanowicz and Ganczarczyk (1983). The correlation gave the lowest mean deviation for the data of Ngian and Martin (1960) and for the data of Timmermans and van Haute (1984). But it gave a higher mean deviation than the correlations of Wen and Yu (1966) and Richardson and Zaki (1954) for this study data. For the total data, it had a mean deviation of 9.2%. In contrast, the Setiadi's (1989) correlation had a slightly higher mean deviation than those of Hermanowicz and Ganczarczyk (1983) for the data of Ngian and Martin (1980) and Timmermans and van Haute (1984). But it predicted well the bed voidage of anaerobic fluidised-bed experiments. Furthermore, the correlation of Setiadi (1989) gave the lowest mean deviation for the total data with a number of data of 160 which was 7.5%. Table 4. Comparison of experimental and calculated value different correlations of bed voidage for different set of data Mean Deviation (%) Experimental data of

Correlations 1. Richardson ~ and Zaki (1954) 2. Wen and Yu (1966) 3.lIermanowicz and Ganczarczyk (1983) 4. Mulcahy and Shieh (1987) S. Setiadi (1989)

Ngian & Martin (1980)

Timmermans This study & van Haute (1984)

Total data

7.6

18.0 19.5

11.5 5.0

13.5

5.9 22.3 6.9

13.8 49.6 14.1

11.6 25.7 1.7

9.2 29.9

12.1

10.1

7.5

Predicting bed expansion of an anaerobic fluidised-bed bioreactor

191

In summary, the correlation of Mulcahy and Shieh (1987) predicted much higher bed voidage than those experimentally measured. The spherical correlation of Richardson and Zaki (1954) always gave a mean deviation of more than 10%. In contrast the other correlations predicted the experimental data reasonably well, since the mean deviations for total data were always less than 10.5%. However, the correlation of Wen and Yu (1966) had a high mean deviation for Timmermans and van Haute (1984), it was just below 20%. The correlation of Hermanowicz and Ganczarczyk (1983) and Setiadi (1989) had a mean deviation for the total data less than 9.5% with the mean deviation for every set of data less than 15%. However, the correlation of Setiadi (1989) predicted the bed voidage of anaerobic fluidised-bed bioreactors with better accuracy than any previously published correlations.

CONCLUSIONS 1. The wet density of biofilm of anaerobic fluidised-bed for wastewater treatment was found to be 1.09 g1cm 3• It was not much different from that of biofilm in a denitrification (anoxic) system. 2. The correlation of Setiadi (1989) predicted the bed expansion of anaerobic fluidised-bed bioreactors with better accuracy than any previously published correlations. ACKNOWLEDGEMENT The author wishes to acknowledge Dr. V. F. Larsen from the University of Strathclyde, Glasgow, UK for his guidance and advice throughout the course of the study. The author is also grateful to the Commitee of ViceChancellors and Principals of the Universities of the United Kingdom, the University of Strathclyde, and the Indonesian Government for financial support. REFERENCES Chen, S. J., Li C. T., Shieh, W. K. (1985). Performance evaluation of the anaerobic fluidized bed system: I. Substrate utilization and gas production. J. Chem. Tech. Biotechnol., 358, 101·109. Cleasby, J. L., Fan, K. S. (1981). Predicting fluidization and expansion ofmter media. J. Enll. Eng Dill, ASCE. 107, No. EEJ, 455471. Davison, B. H. (1988). Dispersion and holdup in a three phase fluidized bed bioreactor. Proc. 10th Symposium on Biotechnology for Fuels and Chemicals, Gatlinburg, Tennessee, May 16-20. Hennanowicz, S. W., Ganczarczyk, J. J. (1983). Some fluidization characteristics of biological beds. Biottchnol. Bioeng.. 25, 1321·1330. lza, J. (1991). Fluidized bed reactors for anaerobic wastewater treatmenL Wat. Sci. Tech., 24, 109-132. Mulcahy, L. T., Shieh, W. K. (1987). Fluidization and reactor biomass characteristics of the denitrification fluidized bed biofiIm reactor. Wat. Res., 21, 451458. Ngian,·K. F., Martin, W. R. B. (1980). Bed expansion characteristics of liquid fluidized particles with attached mierobial growth. Riotec/moL Bioeng.. 22, 1843·1856. Richardson, J. F., Z1kl, w. N. (1954). Sedimentation and fluidisation part I. Trans. Instn. Chern. Eng., 32, 35-53. Setilld~ T. (1989). liquid mixinR and bed expansion in a jluidised·bed reactor with particular reference to gas producing anaerobic systems., PbD.thesis, University of StraUlclyde, Glasgow. Shieh. W. K., Sutton, P. M.. Kos, P. (1981). Predicting reactor biomass concentration in a fluidized system, J.WPCF, 53, 1574· 1584. Tunmenn:ms, p.. van Haute, A. (1984). Biomass hold up in a fluidized bed reactor. App. MicrobiaL Biotech-nol.. 19,3643. Wen, C. Y., Yu, Y. H. (1966). Mechanics of fluidization. Chem. Eng. Progress Symp. Series, 62,100-111.