Influence of organic loading rate and hydraulic retention time

on the efficiency of a UASB bioreactor treating a canning

factory effluent

W Trnovec and TJ Britz*

Department of Food Science, University of Stellenbosch, Private Bag X1, Matieland 7602, South Africa

Abstract

A mesophylic laboratory-scale upflow anaerobic sludge bed bioreactor design was evaluated for the treatment of a carbohydrate-
rich effluent from the canning industry. The bioreactor was inoculated with 500 g of anaerobic granules and after the system
had stabilised the hydraulic retention time (HRT) was set at 24 h and the substrate pH poised at 8.0 to prevent the effect of
rapid acidification. In the first experimental study the chemical oxygen demand (COD) was increased stepwise from 2 300
to a full strength of 4 000 mg·l-1. In the second study the organic loading rate was increased by shortening the HRT (24 to 8 h) to
give an organic loading rate increase from 3.95 to 10.95 kgCOD·m-3·d-1 with an average COD removal of 90 to 93% and
removal rate of 9.8 kgCOD·m-3·d-1. However, the recovery rate of the system at HTR values below 10 h was found to be very
slow suggesting that the system had reached its minimum HRT. This was confirmed by the stabilisation of the granule bed. An
HRT of 10 h was thus taken as the optimum operational HRT. Since neutralisation costs would influence economic aspects of
the process, the influence of lower pH values was investigated in the third study where the pH of the canning effluent was
lowered from 8.0 to 5.0. At the lower pH the COD removal dropped drastically, the biogas production decreased and the
digester effluent pH dropped to 6.2. It was clear from the slow recovery of the digester and the low COD removal (66.1%) that
the lower end of the operational pH had been reached and any further lowering of the substrate pH would lead to system failure.
The economic implication of being able to operate at pH 5.5 means that fresh canning effluent can be introduced into the
digester without any neutralisation, is considerable.

upflow anaerobic sludge blanket (UASB) reactor has become one

Introduction of the most popular designs for the biological treatment of
effluents, and in particular those from the food processing
All industries are increasingly required to reduce their impact on industries (Lettinga et al., 1997). Many UASB reactors are in
the environment. Adequate treatment of food processing efflu- operation throughout the world (Schmidt and Ahring, 1996). The
ents is assuming increasing importance, as this industry ad- advantage of the UASB design is the ability to retain high biomass
dresses the issue of responsible environmental management concentrations despite the upflow velocity of the waste water and
(Wayman, 1996). For the food industry this is frequently difficult the production of biogas. Consequently, the reactor can operate
as factors related to seasonal operation, changes in plant effluent at short hydraulic retention times since the sludge retention time
characteristics due to the processing of different products, nutri- is almost independent of the hydraulic retention time. In UASB
tional deficiencies and the location of food processing plants, reactors, the biomass is retained as granules, formed by the
heavily impact the treatment efficiency. Furthermore, the grow- natural self-immobilisation of the bacteria. These granules have
ing concern over the quality and quantity of freshwater has forced good settling abilities and vary in size from 0.14 to 5 mm
higher surcharges and fines in an attempt to reduce the pollution depending upon the waste water used and operational conditions.
loading on treatment facilities and environmental pollution. The granules vary widely in shape, but they usually have a
Many local authorities are now insisting that industries undertake spherical form. The development of granular sludge is the key
some form of effluent treatment so as to protect the environment. factor for successful operation of UASB reactors. Today it is
Considerable interest has been shown in the application of possible to develop granules on a variety of waste waters and
anaerobic digestion to waste waters from the food industry since defined media, but there have been several reports on lack of
the nature and strength of the waste waters often provide the ideal granulation on specific waste waters (Wentzel et al., 1994).
conditions for digester operation. The waste waters have a high Furthermore, some researchers have reported sudden disintegra-
organic content, have little or no toxic material present (Kroyer, tion of granules without any obvious reason.
1995) and include the situation where waste waters are produced The objective of this study was to assess the effects of
over a short period of the year such as in the canning industry. shortening the hydraulic retention times on the overall perform-
Anaerobic processes have been shown to be amenable to such ance and stability of an UASB reactor, while treating a canning
variations and in particular where complete shutdown may take factory effluent.
place.
Among the high-rate anaerobic reactors developed and suc- Materials and methods
cessfully applied in recent years (Lettinga et al., 1997), the
Digester design

*To whom all correspondence should be addressed. A laboratory-scale upflow anaerobic sludge blanket bioreactor
( (021) 808-3509; fax (021) 808-3510; e-mail [email protected] (UASB) was used. The digester had an operational volume of
Received 6 August 1997; accepted in revised form 5 January 1998 2.3 l (total height of 830 mm and internal dia. of 50 mm) and

ISSN 0378-4738 = Water SA Vol. 24 No. 2 April 1998 1147

 with calcium hydroxide to optimise the environment for maxi-
mum granule growth. This substrate was then diluted to the
required COD concentration.
The effluent from the fruit canning factory was sampled over
the whole 1996/1997 canning season from the waste-water stream
before it reached the general stream that also contained the lye
effluents (Visser, 1997).

Analytical methods

The following parameters were monitored according to the

APHA (Standard Methods, 1985): pH; alkalinity; total solids
(TS); total volatile solids (TVS); and total non-volatile solids
(TNVS). COD, orthophosphate phosphorus and total Kjeldahl
nitrogen were determined colorimetrically using a DR2000 spec-
trophotometer (Hach Co. Loveland, CO) and standardised proce-
dures (Standard Methods, 1985). The general mineral analyses
were done colorimetrically according to standard Hach proce-
dures using a DR2000 spectrophotometer (Hach Co. Loveland,
CO).
The total volatile fatty acids (TVFA) were determined using
a Varian (Model 3700) gas chromatograph, equipped with a
flame ionisation detector and a 30 m x 0.32 mm i.d. Fused Silica
capillary column with 007 FFAP bonded phase (Quadrex Co.
New Haven). The column temperature was initially held at 105°C
for 5 min, then increased at a rate of 10°C min-1 to 219°C. The
detector and inlet temperatures were set at 260°C and 250°C
respectively and nitrogen gas was used as carrier gas at a flow rate
Figure 1 of 2.5 ml·min-1.
Laboratory-scale UASB bioreactor The biogas composition was determined on a Fisons GC
equipped with a thermal conductivity detector and 2.0 m x 3.0
combined a UASB design with an open gas/solids separator at the mm i.d. column packed with Porapak Q (Waters Ass. Inc,
top of the bioreactor (Fig. 1). The biogas exited through the top, Milford, MA), 80/100 mesh. The oven temperature was set at
while the substrate was introduced into the bioreactor at the base. 55°C and helium was used as carrier gas at a flow rate of 40
The overflow of the bioreactor emptied through a U-shaped tube ml·min-1.
to prevent any atmospheric oxygen from entering the system. Carbohydrate composition was determined on a Dani GC
The upflow velocity within the reactor was set at 2 m·h-1. The equipped with a 2.0 m x 3.0 mm column packed with 1% OV-1
temperature of the insulated bioreactor was maintained at 35°C (Waters Ass. Inc, Milford, MA). The column temperature was
using a heating tape and an electronic control unit (Meyer et al., initially held at 160°C, then increased at a rate of 5°C min-1 to
1985). The volume of the biogas was determined using a 193°C. The detector and the inlet temperatures were set at 250°C
manometric unit equipped with an electronically controlled and 280°C respectively and nitrogen gas was used as carrier gas
counter and a gas-tight valve and the volumes corrected to at a flow rate of 30 ml·min-1.
standard temperature and pressure. The substrate was fed semi-
continuously to the bioreactor by means of a peristaltic pump Experimental studies
(Watson-Marlow 101) controlled by an electronic timer.
The study comprised three experimental studies (I to III). In the
Bioreactor start-up first study (I), the substrate COD concentration was increased
stepwise from 2 300 to 4 000 mg·l-1 in 7 steps. In the second study
The bioreactor was seeded with 500 g of water drained anaerobic (II), the COD concentration was kept constant at 4 000 mg·l-1,
granules from another anaerobic digester giving a settled sludge- while the HRT was reduced stepwise from 24 to 8 h in 14 steps.
bed height of 300 mm. The bioreactor was then allowed to In the third study (III), the HRT was reset at 10 h while the COD
stabilise for 48 h in order to allow the bacterial community concentration was kept constant at 4 000 mg·l-1. The substrate pH
to acclimatise and fed with a diluted synthetic substrate (2 300 was then reduced stepwise from 8.5 to 5.0 in 7 steps. In all three
mg·l-1 COD) and the HRT set at 24 h. After three weeks this was studies, the bioreactor was allowed to reach stable-state condi-
replaced with a fruit-canning factory effluent and the COD tions before each HRT or pH reduction. Stable-state is defined as
concentration was gradually increased to 4 000 mg·l-1. a state which can be maintained indefinitely without system
failure (Cobb and Hill, 1990), during which the variation in
Substrate bioreactor performance parameters is less than 10%. Thus, the
length of each phase was based on the stability of the bioreactor
The composition of the synthetic substrate (in mg·l-1) was: effluent pH, alkalinity and COD removal.
glucose 1 250; sodium lactate 5 000; acetic acid 10; urea, 500; and
K2HPO4 500. The substrate was also supplemented with 1.0 ml
trace element solution (Nel et al., 1985) and the pH poised at 8.5

1148 ISSN 0378-4738 = Water SA Vol. 24 No. 2 April 1998

Results and discussion Study I - Increasing the organic loading rate

Canning effluent composition The substrates used during these experimental studies were a
dilution of the factory effluent given in Table 1. To prevent a
The average composition of 15 different batches from a local shortage of nitrogen and phosphorus, a 100 mg·l-1 of each of
fruit-canning factory is given in Table 1. The data clearly show ureum and K2HPO4 and 1.0 ml·l-1 of the trace element solution
that the composition of the effluent was fairly constant over the were added and the COD diluted to the required concentration.
whole canning season. During the study period only the COD of The pH was poised at 8.0 using a 1.0 N Ca(OH)2 solution as
this effluent was standardised. initially at the start of the study the digester showed signs of pH
instability with a tendency towards pH values below 6.5 units. A
summary of the operational conditions and digester efficiency is
TABLE 1 given in Table 2. The hydraulic retention time was kept constant
AVERAGE COMPOSITION OF THE FRUIT CANNING at 24 h and the OLR increased in 7 steps from 2.28 to 3.95
FACTORY EFFLUENT USED AS SUBSTRATE FOR kgCOD·m-3·d-1. The final value represented on average the full
THE UASB BIOREACTOR (mg·l-1) strength effluent as obtained from the factory and from these data
it was concluded that it could be treated directly without dilution.
Parameter Average SD
Study II - Shortening the hydraulic retention time
pH 5.45* 1.4
COD 4432* 297 During this study the HRT was shortened from 24 h to 8 h over
TS 2114* 625 14 steps (Table 3) with a subsequent increase in OLR from 3.95
TVS 1783* 169 to 10.95 kgCOD·m-3·d-1. Stable-state conditions plus 5 HRTs
TNVS 331* 152 were used as criterion for increasing the OLR. During the
PO4 5.32* 3.9 different steps the pH of the digester effluent remained fairly
TKN 21.4* 12.4 constant (7.5 to 8.1) with the alkalinity in the range of 1 800 to
Alkalinity (as CaCO3) 25* 6.5 3 200 at the end of each step. However, directly on changing the
Glucose 161.6+ 25.3 OLR it was usually found that the pH dropped from 0.2 to 0.8 units
Fructose 389.7+ 34.6 but within 5 d the pH increased and stabilised. The pH stability
Sorbitol 88.8+ 9.7 can probably be ascribed to the high alkalinity level (1 800 to
Ca2+ 21+ nd 3 200 mg·l-1). According to Duff and Kennedy (1982) and Lane
Co2+ <0.06+ nd (1984), alkalinity plays an important role in minimising over-
Fe (total) 7.9+ nd loading effects.
K 84+ nd During this study it was found that at HRT values of shorter
Mg 15+ nd than 10 h, the recovery rate in terms of pH and COD removal
Na 65+ nd stabilisation was slower than found with the longer HRTs with up
Ni2+ <0.3+ nd to 14 h before the two parameters stabilised. The decrease in pH
SO42- 12.8+ nd after increasing the OLR, as one of the indicators of impending
S2- 0.03+ nd digester failure, has been intensively studied (Hill and Bolte,
1989). According to Dohanyos et al. (1985), any change in
* = Data are means of 15 batches operational parameters, such as organic loading, causes simulta-
+
= Average of two determinations neous increase in the concentration of all the volatile fatty acids
resulting in a decrease in the pH. Once the microbial biomass has
recovered and stabilised the extra VFAs are normally metabo-
lised and the pH stabilises (Myburg and Britz, 1993). Based on
the extended stabilisation time at these HRTs, it was concluded

TABLE 2
OPERATING CONDITIONS AND DIGESTER EFFICIENCY DURING EXPERIMENTAL STUDY I WHERE THE
ORGANIC LOADING RATE WAS INCREASED

Parameter Steps

1 2 3 4 5 6 7

Substrate COD (mg·l-1) 2 300 2 500 3 000 3 200 3 500 3 700 4 000
COD removal (%) 88 92 90 91 89 90 91
HRT (h) 24 24 24 24 24 24 24
OLR (kgCOD·m-3·d-1) 2.28 2.5 3.0 3.22 3.49 3.73 3.95
Digester pH 7.6 7.5 7.3 7.4 7.4 7.7 8.0
Alkalinity (mg·l-1 CaCO3) 1 800 1 850 1 950 1 125 1 225 2 010 2 125
Biogas (l·d-1) 1.1 1.37 2.36 1.84 2.27 2.35 2.52
Methane (%) 62 63 64 64 65 64 64

ISSN 0378-4738 = Water SA Vol. 24 No. 2 April 1998 1149

 that even though the COD removal after stable-state had been
reached was still above 90%, the digester was reaching its
maximum operational HRT and that any sudden changes in
normal operating parameters would influence the efficiency
negatively. An HRT of 10 h with an average COD removal of
between 90 and 93% and removal rate of 9.8 kgCOD·m-3·d-1 was

4 161

3 200
10.95

8.7
7.9
93

64
14
taken as the optimum operational conditions (Table 3).

8
Study III - Lowering of the substrate pH

4 099

2 700
6.68
7.9
9.8
91

64
13

9
OPERATING CONDITIONS AND DIGESTER EFFICIENCY DURING EXPERIMENTAL STUDY II WHERE THE HRT WAS SHORTENED

Since neutralisation costs would influence economic aspects of

the process, the influence of lower pH values was investigated in
3 991

2 350
5.72
8.68
7.6
the third experimental study. In this study, based on the data
90

62
10
12

obtained during Study II, the HRT was kept constant at 10 h and
the substrate pH lowered in 7 steps over 60 d from 8.0 to 5.0.
4 125

2 625
6.39
8.25

The data (Table 4) show that at substrate pHs of 6.0 and 5.5,
8.1
93

64
11
11

the alkalinity drastically decreased to 1 150 mg·l-1. However, the

COD removal was still above 88% and the effluent pH above 6.8
4 069

2 450

which is still in the optimal recommended pH range (Nel and

6.14
7.54
8.1
93

64
12
10

Britz, 1986). When the substrate pH was lowered to 5.0 the COD
removal dropped drastically (Table 4) and the biogas production
3 967

2 600

started to decrease. It was also found that just after the change to
7.55

5.58
6.84
91

64
13
9

pH 5.0 the digester effluent pH dropped to 6.2 and once the system
reached stable state slowly increased to 6.7. It was clear from the
slow recovery of the digester and the low COD removal (66.1%)
3 953

3 800
5.33
6.27
7.7
90

63
14
8

that the lower end of the operational pH had been reached and any
further lowering of the substrate pH would lead to system failure.
The substrate pH was then reset at 5.5 which is near the average
Steps

3 912

3 200
4.5
7.5
6.2
88

62
15

pH of the fresh raw canning effluent (Table 1) and the COD

7

removal and digester effluent pH slowly recovered to about 90%

TABLE 3

COD removal and pH 6.8 to 7.0. The economic implication of

4 233

2 200
4.72
6.13
7.6
94

65
16

being able to optimise and operate the digester at a substrate pH

6

of 5.5 is considerable since it means that fresh canning effluent

can be introduced into the digester without any neutralisation. It
4 119

2 250
4.7
8.1
5.6

must also be remembered that the canning effluent is rich in

92

64
17
5

carbohydrates and if the effluent is stored, fermentation will take

place with a concurrent reduction in pH to values of about 3.0 to
3 932

1 800
4.74

3.5. Neutralisation must then be applied before substrate intro-

4.4
7.9
93

65
18
4

duction to the digester.

4 011

2 100

UASB bioreactor efficiency

2.78
4.36
7.5
96

64
20
3

Since the efficiency of the UASB system is based on the forma-

4 162

2 250

tion and retention of granules, changes in the height of the UASB

2.29
4.16
7.9
96

64
22
2

granule bed were monitored during the course of the study

(Fig. 2). With a start bed height of 300 mm, growth at first was
very slow but as soon as the HRT was lowered to below 24 h, a
3 947

2 125
2.52
3.95
8.0
91

63
24
1

definite increase in the bed height was found, suggesting that the
bioreactor had not reached its loading capacity. Around day 90,
once an HRT of 10 h had been reached, the bed growth stabilised
and remained stable for the rest of the study. At HRT values
Alkalinity (mg·l-1 CaCO3)

below 10 h the system was found to show an extended recovery

Substrate COD (mg·l-1)

OLR (kgCOD·m-3·d-1)

time indicating maximum loading at the shorter HRTs. It is

COD removal (%)

possible that if the COD concentration of the canning effluent

itself had been higher, a further increase in granule bed height
Methane (%)
Biogas (l·d-1)
Digester pH

might have taken place but since 4 000 mg·l-1 was the maximum
Parameter

concentration, this was not followed up.

HRT (h)

The UASB bioreactor efficiency in terms of the relationship

between percentage COD removal and the COD removal rate
(kg COD·m3·d-1) is plotted in Fig. 3 as a function of the OLR over
all three the experimental studies. The data obtained at HRT
values below 10 h (Table 3) and those at a pH value below 5.5
(Table 4) were not included as the slow bioreactor recovery

1150 ISSN 0378-4738 = Water SA Vol. 24 No. 2 April 1998

 TABLE 4
OPERATING CONDITIONS AND DIGESTER EFFICIENCY DURING EXPERIMENTAL STUDY III WHERE THE
SUBSTRATE pH WAS DECREASED

Parameter Steps

1 2 3 4 5 6 7

Substrate COD (mg·l-1) 3 937 3 980 4 096 3 834 4 018 4 104 4 352
COD removal (%) 90.7 88.5 90.5 90.2 88.3 89.4 66.1
HRT (h) 10 10 10 10 10 10 10
OLR (kgCOD·m-3·d-1) 8.75 9.48 9.75 9.13 9.57 9.77 10.4
Substrate pH 8.0 7.5 7.3 6.5 6.0 5.5 5.0
Digester pH 8.0 7.6 7.89 7.95 6.82 6.91 6.7
Alkalinity (mg·l-1 CaCO3) 3 475 2 550 2 775 2 175 1 150 1 150 1 050
Biogas (l·d-1) 4.88 5.72 6.45 7.7 8.75 9.43 8.12
Methane (%) 64 63 63 64 65 65 63

clearly indicated that the lower end of the operational

HRT and pH had been reached and any further changes
in OLR would lead to system failure. The best removal
rate (R-value) was found in the OLR region of about 11
kgCOD·m-3·d-1 (Fig. 3). Since the highest COD concen-
tration (4 000 mg·l-1) was already being loaded, any
increase in OLR could only be obtained by further
lowering of the HRT to below 10 h. However, changes
to lower HRT values led to a slow recovery of the
bioreactor suggesting that the system had reached its
maximum organic loading rate and that system failure
would be imminent even after small environmental
changes (Verstraete and Vandevivere, 1997).

Conclusions

It has long been thought that anaerobic digestion is too

slow and unreliable to be used by the canning industry
Figure 2 as a treatment option (Borja and Banks, 1994; Wayman,
Changes in the height of the granule bed in the UASB 1996). From the results obtained during the three experi-
mental studies it was clear that the UASB design is
feasible for treatment of the carbohydrate-rich effluents
produced in the canning industry. The most favourable
COD removal of the canning-industry effluent was
between 89 and 93% at organic loadings of 9.8 and 10.95
kgCOD·m-3·d-1, at an HRT of 10 h and substrate pH of
5.5. The UASB bioreactor in terms of HRTs, OLRs and
substrate pH as operated in this study, was more effi-
cient when compared to results reported by Austermann-
Hann et al. (1997), where a UASB was used to treat a
fruit-juice effluent.

Acknowledgements

The financial support of the Water Research Commis-

sion and the University of Stellenbosch is gratefully
acknowledged. We would also like to acknowledge the
co-operation of Mr J Visser of Ashton Cannery in
suppling the cannery effluent and Dr A Wood for his
help in obtaining a regular supply of granules.
Figure 3
The effect of the increase in organic loading rate on the percentage
COD removal (¡) and the COD removal rate (l)

ISSN 0378-4738 = Water SA Vol. 24 No. 2 April 1998 1151

References MEYER LH, HUGO AB, BRITZ TJ, DE WITT B and LATEGAN
PM (1983) Temperature control for laboratory scale anaerobic
AUSTERMANN-HANN U, SEFRIED CF and ROSENWINKEL K digesters. Water SA 9 (2) 79-80.
(1997) UASB-reactor in the fruit juice industry. Proc. 8th Int. Conf. MYBURG C and BRITZ TJ (1993) Influence of higher organic loading
Anaerobic Digestion (Vol 1.) Sendai, Japan. 67-74. rates on the efficiency of an anaerobic hybrid digester while treating
BORJA R and BANKS CJ (1994) Kinetic study of anaerobic digestion landfill leachate. Water SA 19 (4)319-324.
of fruit-processing wastewater in immobilized-cell bioreactors. NEL LH, BRITZ TJ and LATEGAN PM (1985) The effect of trace
Biotechnol. Appl. Biochem. 20 79-92. elements on the performance efficiency of an anaerobic fixed film
DOHANYOS M, KOSOVA B, ZABRANSKA J and GRAU P (1985) reactor treating a petrochemical effluent. Water SA 11 (3)107-110.
Production and utilization of volatile fatty acids in various types of NEL LH and BRITZ TJ (1986) The influence of different pH values on the
anaerobic reactors. Water Sci. Technol. 17 191-205. performance of a downflow anaerobic fixed bed reactor treating a
COBB SA and HILL DT (1990) Using nitrogen ratio as an indicator of petrochemical effluent. Biotechnol. Lett. 8 293-298.
biomass retention and steady state in anaerobic fermentation. Trans. SCHMIDT JE and AHRING BK (1996) Granular sludge formation
ASAE 33 282-287. in upflow anaerobic sludge blanket (UASB) reactors. Biotechnol.
DUFF SJB and KENNEDY KJ (1982) Effect of hydraulic and organic Bioeng. 49 229-246.
overloading on thermophylic downflow stationary fixed film (DSFF) STANDARD METHODS (1985) Standard Methods for the Examination
reactor. Biotechnol. Lett. 4 815-820. of Water and Wastewater (16th edn.) American Public Health Associa-
HILL DT and BOLTE JP (1989) Digester stress as related to iso-butyric and tion, Washington DC.
iso-valeric acids. Biol. Wastes 28 33-37. VERSTRAETE W and VANDEVIVERE P (1997) Broader and newer
KROYER GT (1995) Impact of food processing on the environment - An applications of anaerobic digestion. Proc. 8th Int. Conf. on Anaerobic
overview. J. Food Sci. 28 547-552. Digestion (Vol 1.) Sendai, Japan. 67-74.
LANE AG (1984) Laboratory scale anaerobic digestion of fruit and VISSER J (1997) Personal communication. Ashton Canning Co. (Pty) Ltd.
vegetables solid waste. Biomass 5 245-259. WAYMAN MJV (1996) Water supplies, effluent disposal and other envi-
LETTINGA G, HULSHOFF POL LW, ZEEMAN G, FIELD J, VAN ronmental considerations. In: D Arthey and PR Ashurst (eds.)
LIER JB, VAN BUUREN JCL, JANSSEN AJH and LENS P (1997) Fruit Processing. Blackie Academic & Professional Press, London.
Anaerobic treatment in sustainable environmental production con- 221-243
cepts. Proc. 8th Int. Conf. Anaerobic Digestion (Vol 1.) Sendai, Japan. WENTZEL MC, MOOSBRUGGER RE, SAM-SOON PALNS, EKAMA
32-39. GA and MARAIS GvR (1994) Tentative guidelines for waste selec-
tion, process design, operation and control of upflow anaerobic sludge
bed reactors. Water Sci. Technol. 30 31-42.

1152 ISSN 0378-4738 = Water SA Vol. 24 No. 2 April 1998