Sie sind auf Seite 1von 11

Journal of Biotechnology 45 (I 996) I25- 135

Anaerobic treatment of palm oil mill effluent in a two-stage


up-flow anaerobic sludge blanket (UASB) system
Rafael Borja ap*, Charles J. Banks b, Enrique Sinchez ’
a Institute de lu Grusa (C.S.I.C.). Audu. Padre Gurciu Tejero 4, E-41012 Seuillu, Spin
b Enuironmentul Technology Cenrre, Department ofChemicuf Engineering. Lfniversizy @‘Manchester Institute #‘Science and Technology
(U.M.I.S.T.), P.O. Box 88, Manchester M60 IQD, UK
’ Depurtamento de Estudios de Contaminu&n Ambiental (D.E.CA., C.N.I.C.), P.O. Box 6990, Lo Hubanu. Cuhu

Received 6 June 1995; reviskd 21 July 1995; accepted 11October 1995

Abstract

A high rate anaerobic treatment of palm oil mill effluent (POME) was achieved in a two-stage up-flow anaerobic sludge
blanket (UASB) reactor. The acidogenic reactor acclimated rapidly to the wastewater and was tolerant to a suspended solids
(SS) concentration of 5.4 g 1-r in the influent wastewater. Loading was gradually increased over a period of 100 days
resulting in a satisfactory hydrolysis and acidification giving a maximum rate of acid production of 4.1 g I- ’ d _ ’ acetic acid
at a loading rate of 16.6 g I-’ d-’ COD at a hydraulic retention time of 0.9 days. An increase in alkalinity throughout the
acclimatization maintained the effluent from the reactor at around pH 5.8. The methanogenic reactor was initially fed on
dilutions of the effluent from the first stage reactor after pH adjustment. The loading was gradually increased, and then
stepwise, to 60 g I-’ dd’ COD at which point COD removal efficiency had declined significantly and an accumulation of
long-chain volatile fatty acids was observed. It was concluded that the reactor could work efficiently up to loadings of 30 g
l- ’ dd’ COD, whilst producing a good methane yield and a COD reduction of greater than 90%. Effluent recirculation
alleviated the need for alkali additions to the feed of the methanogenic reactor and a direct coupling of the two reactors was
achieved towards the end of the experimental run of 175 days. Both reactors showed granule formation with distinct
morphological characteristics; these were observed to be formed after 80 days in the acidogenic reactor and after I IO days in
the methanogenic reactor.

Keywordx Palm oil mill effluent (POME); Anaerobic treatment; Two-stage UASB system; pH Control

1. Introduction been given in detail elsewhere (Borja and Banks,


1994).
The three main sources of liquid effluent gener- For a well-controlled conventional mill, about 0.9,
ated from a conventional palm oil mill are sterilizer 0.1 and 1.5 m3 of steriliser condensate, separator
condensate, hydrocyclone waste and separator sludge. sludge and hydrocyclone waste are generated for
The process flow-sheet for palm oil extraction has each tonne of crude palm oil produced. In most
mills, these three wastewater streams are combined
together resulting in a viscous brown liquid contain-
* Corresponding author. ing fine suspended solids, the characteristics of which

0168- 16.56/96/$15.00 0 1996 Elsevier Science B.V. All rights reserved


SSDI 0168-1656(95)00154-9
126 R. Borju et aI./ Journal of’Biotechnology 45 (1996) 125-135

Table I retention time but ran into operational problems of


Properties of palm oil mill effluent (POME) poor sludge settleability. Chin (1981) described a
Parameter a Sterilizer Hydrocylone Centrifuge Mixed series of bench-scaie experiments comparing full
effluent effluent effluent effluent
mixed continuous stirred tank reactor (CSTR) di-
BOD IO-25 - 17-35 I l-30 gesters with similar digesters operating with a cell
COD 30-60 - 40-75 30-70
recycle system; for a 90% COD removal a hydraulic
Total solids 40-50 5-15 35-70 30-65
5-12 12-18 9-25
retention time (HRT) of 20 d was required. Ibrahim
Suspended 3-5
solids et al. (1984) showed that at 45°C an anaerobic
Oil 2-3 l-5 5-15 5-13 contact digester could be operated at a maximum
Ammoniacal 0.02-0.08 - 0.02-0.08 0.02-0.08 organic load of 6.2 g 1-l d-’ COD, giving a per-
N centage reduction of around 94%; a higher loading
Total N 0.35-0.60 0.07-0.15 OS-O.9 0.5-0.9 of 7.0 g 1-l dd ’ could be treated but only at the
PH 4.5-5.5 - 3.5-4.5 3.5-4.5
higher temperature of 50°C. Cail and Bat-ford (1985)
a In g I-‘. used a semicontinuous mesophilic anaerobic digester
with a volumetric loading of 12.6 g I-’ d- ’ (HRT of
are shown in Table 1. If discharged untreated this 6 d). Ng et al. (1985) showed 85% reduction in
effluent can cause considerable environmental prob- pollution load using a two-stage anaerobic digestion
lems (Davis and Reilly, 1980). system with no cell recycle and a total HRT of 31 d.
The increasingly stringent water quality regula- The data of Edewor (1986) showed that a IO-day
tions being introduced in many countries have forced retention time in a single-stage anaerobic contact
factories to investigate a wide range of approaches digester gave a COD removal efficiency of 93.8%. A
for the treatment of palm oil mill effluent (POME) two-stage variant of this system gave an average
and related wastewaters (Petitpierre, 1982). These removal efficiency of about 96% for HRTs between
include: simple skimming devices (Roge and Ve- 7 and 15 d. The use of an up-flow anaerobic sludge
layuthan, 1981; Ng et al., 1988); land disposal (Ma blanket (UASB) reactor for treating this wastewater
and Ong, 1986); use as animal fodder (Sutanto, has been reported in a previous paper (Borja and
1983; Ma and Ong, 1986); ultrafiltration (Suwandi, Banks, 1994). Over 80% COD was removed at
1981; Turisin et al., 1981); chemical coagulation and loading rates up to 10.6 g 1-l d-’ COD, and at the
flotation (Badri, 1984; Ho and Chan, 1986; Chin et highest influent concentration (42.5 g 1-l > reactor
al., 1987; Ng et al., 1987a; Karim and Hie, 1987); instability was observed.
and various aerobic (Abdul et al., 1989) and anaero- Although the UASB reactor has been successfully
bic microbiological processes. Anaerobic biological applied to a wide range of both low and medium
systems offer greater potential for the treatment of strength food industry wastewaters the mechanism of
POME, as they do not have the high energy demand bacterial granulation is still not fully understood
of aeration as required by aerobic biological systems (Lettinga et al., 1980; Lettinga and Vinken, 1980;
(Edewor, 1986; Ng et al., 1987b; Ma and Ong, Goodwin et al., 1992). The process appears to be
1988). The use of conventional anaerobic lagoons particularly sensitive to the suspended solids (SS)
(Hum et al., 1981; John, 198.5) or digesters (Ho and loading which can be applied, and also to the nature
Tan, 1985) to treat POME is characterized by long of the organic content of the wastewater. Souza
residence times, often in excess of 20 d, necessitat- (1986) reported that for a single-stage UASB reactor
ing large areas of land or large digesters. The appli- then a value of 0.04 g SS g- ’ COD could be treated
cation of modem high rate anaerobic digester tech- and provided the ratio was not exceeded then sus-
nologies (Pandey and Ramakrishna, 19901, such as pended solids concentrations as high as 1 g 1-I could
up-flow or down-flow filters, fluidized beds, up-flow be applied.
anaerobic sludge blanket (UASB) systems or up-flow At high organic loadings or at low temperatures
floe digesters, for the disposal of POME is rare. the insoluble organic fraction of the wastewater tends
Peyton et al. (1979) used a laboratory scale to accumulate within the granule or sludge blanket
mesophilic contact stabilization process with a IO-day region of the reactor; this leads to granule destabi-
R. Borju et ul./JournuI of Biorechnology 45 (1996) 125-135 127

lization or inhibition of granule formation (Hulshoff UASB reactor (A) with 13 cm i.d. and a 5-1 UASB
Pol and Lettinga, 1986; Vigneswaran et al., 1986). reactor (MI with 9 cm i.d. were used for the acido-
Under these conditions methanogenesis is affected, genie and methanogenic reactions of a two-stage
leading ultimately to reactor failure as a result of process. The reactors were separated by a 2.5-l bal-
increased acid concentrations. One solution is to ancing tank receiving the effluent from reactor A and
remove the solids from the wastewater stream prior providing the feed reservoir for reactor M. The
to it entering the reactor as was the case in previous reactors and balancing tank were maintained at a
studies using POME. This is now, however, ideal for constant temperature of 35°C in an insulated cabinet.
two reasons. Firstly, the suspended organic solids Each reactor was of a typical UASB design having a
have a high potential for gas production making gas-biomass-liquid separator at the head of the
process economics more favourable. Secondly, re- column and an influent liquid distributor at the base.
moval prior to UASB treatment generates a sludge Each column was connected to a water displacement
which requires disposal at additional cost. type gasometer filled with an acidified brine solution
Hydrolysis and acidification of organic solids in a to prevent CO, dissolution. The reactor contents
separate stage to that of methanogenesis was first could be sampled at various heights along the col-
proposed by Ghosh et al. (1975) and two-stage reac- umn by means of a series of six sampling ports along
tor systems, using conventional continuous stirred its length. A detailed design description of this type
tank reactors (CSTRs), are not uncommon. To apply of small-scale UASB reactor has been given else-
the principal of a two-stage treatment using a pair of where (Borja and Banks, 1994).
upflow anaerobic reactors is the subject of this re-
search, the objective being to evaluate the effect of a 2.2. Wastewater
staged treatment on sludge granulation.
The reactors were fed on dilutions of a mixed
waste stream (POME) typical of that produced in the
2. Materials and methods palm oil manufacturing process. The features and
composition of the POME are summarized in Table
2.1. Experimental design 2, the data shows average values from five separate
determinations for each parameter; the deviation be-
The schematic diagram of the two-stage UASB tween these repetitive analyses was less than 3%.
system used in this study is shown in Fig. 1. A 12-l During the initial feeding of the wastewater to the

EPFLUENT

RECIRCULATION

A: Acidogenic UASB reactor. S: Separation tank.

M: Methanogenic UASB reactor. P: Periststtic pump.

R: Refrigerated feed tank. G: Gas collector.

Fig. I. Schematic diagram of the two-stage UASB system used in this study.
128 R. Borju er ai./Journal oj’Biorechn&gy 45 (19961 125-135

Table 2 the first 100 days of operation an influent SS concen-


Composition and features of the POME used tration of 5.4 g 1-l was maintained by diluting
PH 4.4 POME with tap water. Between days 101 to 120 the
COD(g I-‘) 30.6
influent SS concentration was increased to 10.8 g 1-l
Suspended solids (SS; g l- ’ ) 10.8
8. I by feeding with undiluted POME. From the 121st
Volatile suspended solids (VSS; g 1.. ’ )
Total solids (TS; g I- ’ ) 31.2 day the concentration was dropped once again to 5.4
Volatile solids (VS; g I- ’ ) 24.3 g 1-l SS. The initial loading rate to reactor M was
Kjeldahl N (mg I _ ’ ) 365 1.1 g I-’ d- ’ COD and this was achieved by
P(mg 1-l) 110
dilution with tap water into the balancing tank so as
Fe(mg 1-l) 205
60
to prevent inhibition by volatile fatty acids. As re-
S(mgl-‘1
Ca(mg I-‘) 220 moval efficiency increased above 80% mark the
Na(mg I-‘) 4 loading was gradually increased over a 60-d period
K(mgl-‘) 510 so as to maintain this level of treatment. At the end
Mg(mg 1-l) 170
of this time tap water dilution was no longer neces-
Cu(mgl-‘) I
6
sary but pH adjustment continued until day 140. At
Zn(mgl-‘)
Mo(mp I-‘) 0. I this time alkali additions to the balancing tank were
Co(mg 1-l) 0.007 stopped and after a further 5 d (day 1461, the balanc-
Mn(mgl-‘) 0.6 ing tank was removed and reactor A effluent was fed
Ni(mgl-‘) I .2
directly to reactor M.
Al(mgl-‘1 120
B(mgI-‘) 0.9
In an attempt to prevent a loss in alkalinity from
Ba(mg I-‘) 0.3 the system after alkali additions had ceased, the
Si (mg l- ’ ) 55 effluent from M was recirculated into reactor A from
day 160 onwards.

reactors the pH of the POME was adjusted to 6.0 2.5. Sampling and chemical analyses
using a 5 N sodium hydroxide solution. Different
dilutions of the POME were prepared using tap Gas volume and composition were monitored daily
water so as to cover a range of suspended solids and effluent samples from both reactors were anal-
concentrations. ysed for COD, pH, and VFA (volatile fatty acids)
every 3-4 days; the analytical methods used were in
2.3. Seed sludge accordance with standard methods (American Public
Health Association, 1985).
Reactor A was inoculated with sludge taken from Methane was determined by gas chromatography
a completely mixed acidogenic reactor which was using a stainless steel column (200 X 0.3 cm) packed
part of a two-stage system treating concentrated ice- with active carbon (30-60 mesh), the gas flow pass-
cream wastewater. Reactor M was seeded with ing over a thermal conductivity detector.
methanogenic sludge taken from a CSTR reactor Volatile fatty acids (VFA) were determined using
treating POME. The initial sludge concentrations a gas chromatograph equipped with a 2 m X 4 mm
were 18.8 and 16.1 g 1-l of volatile solids (V’S) for glass column packed with Supelcopor (loo- 120
reactors A and M, respectively. mesh) coated with 10% Fluorad FC 431. The tem-
peratures of the column, the injection port and the
2.4. Loading regime flame ionization detector were 130, 220 and 240°C.
respectively. Nitrogen saturated with formic acid was
Reactor A was started with an initial loading rate used as the carrier gas at a flow rate of 50 ml min - ’ .
of 2.3 g 1-l dd ’ COD and a HRT of 6.5 d. There- The POME was digested in aqua regia, and its
after, the loading rate was increased stepwise up to elemental composition analysed by inductively cou-
17.3 g I- ’ d- ’ COD over a period of 120 d. During pled plasma (ICP) emission spectrophotometry.
R. Borju et d/Journal of Biotechnology 45 (1996) 125-135 129

The structure of the granular sludge was made acetic acid equivalents, can be determined and ex-
with scanning electron microscope @EM). pressed as g acetic acid per 1 reactor volume and day.
As can be seen from Fig. 2, both the rate of acid and
gas production showed a similar trend with increases
3. Results and discussion in both parameters as organic loading increased in
3.1. Stage separation the range 2.5-16.6 g l- ’ d- ’ COD. This corre-
sponds to the time during which the reactor was
Biogas composition and volatile acid concentra- operated with an influent SS concentration of 5.4 g
tion indicated that the stage separation was easily I-‘; the maximum acid production was 4.1 g 1-l
achieved in an early stage. Adjusting pH seems d- ’ acetic acid and corresponded to a HRT of 0.9 d.
simpler than other reported techniques of stage sepa- When the influent SS concentration was increased
ration, such as membrane dialysis, selective inhibi- to 10.8 g 1-l between days 101-120 both the acid
tion and kinetic control (Massey and Pohland, 1978). and gas production rates decreased sharply; this cor-
In addition, a pH of around 6.0 appears to be the responds to an organic loading rate of 17.3 g I-’ dd ’
optimum for acidogenic activity. If the pH is not COD. During this period of operation the height of
controlled the acidogenic biomass tends to buffer the sludge blanket increased and can be attributed to
itself to a pH that depends on the other environmen- an accumulation of non degraded organic solids
tal conditions in the reactor, but mainly on the nature within the reactor. This in turn led to a decreased in
of the wastewater (Zoetemeyer et al., 1982; Hanaki acidogenic activity which may have been as a result
et al., 1987). of the poor mixing observed within the sludge blan-
ket at this time. The gas produced in the acidogenic
3.2. Pelfonance of the acidogenic reactor
reactor was primarily carbon dioxide although at the
From the mass balance of volatile fatty acids in lower organic loadings methane was detected at lev-
reactor A the rate of acid production, in terms of els up to 9%. As the organic loading increased

-1 -1

HAc formation rate (g I d ) Gas production (I d-j


- 18
1
16

14

12
i
10

0
5 15 20
COD loadinrrate (g ;;‘I

L---ff Acid formation rate *- Gas production

Fig. 2. The response of the acidogenic reactor, in terms of the gas (1 d- ’ ) and acid (g I- ’ d- ’ HAc) production rates, to organic loading (g
I-’ d- ’ COD). Th e points at the highest loading value show a depression in rates due to the high influent suspended solids concentration
(10.8 g I-‘).
130 R. Borju et al./ Journal ofBiotechnology 45 (1996) 125-135

methane concentration decreased and hydrogen be- over 4.0 g l-‘, and the proportion of acetic acid to
came detectable and was consistently measured be- that of longer chain acids increased slightly during
tween 1 and 2% from day 80 onwards. The results of this period showing recovery to conditions similar to
the gas analysis are in line with expectations and the those observed shortly before day 100. Propionic,
appearance of hydrogen is indicative of the subse- butyric and valeric acids were all detectable in con-
quent breakdown of longer chain volatile fatty acids centrations greater than 100 ppm. Free VFAs as
to acetic acid and the loss of autotrophic products of the acidogenesis, can diffuse into the
methanogenic activity. During the first 100 days of cells of anaerobic bacteria and ionise to decrease the
operation the reactor alkalinity gradually increased pH. This is generally recognized as inhibiting acido-
from an initial value of 2100 mg l- ’ to a final value genesis (Dinopoulou and Sterrit, 1988). The alkalin-
of 5200 mg 1- ’ as CaCO, at loading rate of 16.6 g ity of the influent allows the neutralization of the
1-l d-’ COD. The pH at the top of the sludge free VFAs, preventing the fall in pH from its opti-
blanket remained relatively constant, between 5.2 to mum value both in the reactant influent and in the
5.8 over this time period showing that the increased cells, while the VFAs in the acidogenic UASB in-
acid production was being satisfactorily compensated creased.
by an increased in alkalinity.
Between days 120- 175 the acidogenic reactor 3.3. Performance of the methanogenic reactor
was fully acclimated and operating with a influent
SS concentration of 3.8 g l-‘, it was however recov- The performance of the methanogenic reactor was
ering from the high loading and solids concentration monitored using the COD removal efficiency (%I,
applied between day 1OO- 120. The predominant VFA the specific gas production rate (1 CH, gg ’ COD
found during this period was acetic acid (Fig. 3) and removed), and the methane content of the biogas
accounted for between 45 to 60% of the total acid (%I; all of these are known to respond to conditions
during this period. The level of total mixed acids of organic overloading and are indicative of the
dropped over a 45-d period from 5.6 g 1-l to just system stability.

VFA concentration (g i’)


61 ~~
1

I I
1” ---- +&_“-_._--_
i’ I
0 --I-27? L i
100 110 120 130 140 150 160 170 180 190 200
Operation time (d)

- c2 c3 - c4 " c5 . Total VFA (


~_
Fig. 3. The volatile fatty acid profile in the acidogemc reactor following the period of maximum loading (days lOI- 119).Total acids are
expressed as acetic acid equivalents (g I- ’ HAc) and individual acids in g I- ’
R. Borja et al./ Journal of Biotechnology 45 (19%) 125-135 131

-1 -1
A (%I; 6 (%I C, biogas (I I d 1; D, COD (g l-b-j
lOOr-
F70

60

1
50

40

30

+$y::2$Q
+-e -9
-~~~ i_ ~__ -L- ___-
20

10

0
50 100 150 200
Operation time (d)
r--
-. A --t B + C --ff D

Fig. 4. Response of the reactor performance indicators in the methanogenic reactor to a gradual, and then stepwise, increase in organic load
over 150 d. The last 25-d period of operation is without alkali additions to the influent and with and effluent recycle to the acidogenic
reactor. (A) Methane content of the biogas (%); (B) COD removal efficiency (%); (C) specific biogas production rate (I I- ’ d- ‘); (D)
organic loading rate (g I- ’ d- ’ COD). -

VFA concentration (g (‘1


4r---------
1

0 i ._~___1__ _._ L

100 110 120 130 140 150 160 170 180 190
Operation time (d)
I

-- c.2 +-C3 --t- c4 ifc5 - Total WA

Fig. 5. Total volatile fatty acid (g I- ’ HAc)and volatile fatty acid components (g I - ’ ) in the methanogenic reactor during the period from
days 120 to 175.
132 R. Borju et al./ Journal c$Biotrchnolo~y 45 (1996) 125-13.5

With any methanogenic reactor it is necessary to rate also rose to a maximum value of 26 1 1-l d- ‘,
assay the biodegradative capacity with respect to whilst the COD removal efficiency showed a steady
increasing organic load so as to optimise reactor size decline to below the 80% level.
whilst maintaining treatment efficiency. To achieve At the same time the volatile fatty acid level in
this the organic load must be either gradually in- the effluent from the reactor rose to 3 and then 3.5 g
creased or increased stepwise over a period of time 1-l mixed acid for the loadings of 45 and 60 g I-’
to allow acclimatization of the microbial population d- ’ COD, respectively (Fig. 5). Of the mixed acids
and pseudo steady state conditions to be reached at present propionic was dominant with levels in excess
each loading value. The reactor showed a rapid of2gl_1, indicating a build up of longer chain acids
initial adjustment to the POME feed reaching 90% and rate limitation in the acidogenic conversion stage.
COD removal efficiency at the initial loading of 1.1 Such conditions normally lead to a rapid souring of
g I- i d- ’ within 10 d (Fig. 4). The increases in load the anaerobic process with eventual inhibition of
applied were of gradual nature over the first 120 methanogenic activity. For this reason the loading
days of operation cumulating in a loading rate of rate was again reduced to 28.2 g 1-l d- ’ COD for
22.9 g I- ’ d- ’ COD. During this period of operation the next 20 d before dropping to a loading of 20.0 g
biogas production increased in line with increases in I-’ d-’ for the final days of the experimental run.
loading, whilst the percentage methane content of the COD removal efficiency improved reaching over
biogas gradually decreased from an initial value of 90%, whilst the gas production rate dropped back to
over 80% to one of around 65%. Although there was 10 I I- ’ d- ‘. The volatile fatty acid levels in the
some variation in COD removal efficiency this aver- reactor effluent rapidly fell to a level of around 0.5 g
aged over 90% with low values on occasions of less I-‘, with the C3, C4, and C5 chain length acids
than 80%, between days 121 and 149 two sharp being removed to concentrations of less than 0.1 g
increases in loading were applied, the first taking the I-‘; this was indicative of a healthy sludge and
load to 45 g 1-l d- ’ COD and the second to 60 g restoration of acetogenic activity. Overall, the
1-l dd ’ COD. During this period the gas production methane yield throughout the course of the experi-

Alkalinity (mg i’)


8000 , ..

6000 i

5000

I--- ! .
4000
140 145 150 155 160 165 170 175 180
Operation time (d)
Fig. 6. Change in alkalinity (mg I ‘, as CaCO,) in the methanogenic reactor on cessation of alkali addition to the influent. An effluent
recycle to the acidogenic reactor was initiated on day 155at a ratio of 0.33 parts reactor M effluent to I part reactor A feed.
R. Borja et cl!./ Journal of Biotechnology 45 (1996) 125-135 133

ment ranged from 0.30-0.33 1 CH, per g COD,,,,,,,, dogenic conditions is far less studied than in mixed
this is a little less than the theoretical value of 0.35. digestion although it has been reported by a number
On day 140 alkali addition to reactor M feed was of authors (Cohen and Zoetemeyer, 1979; Lettinga et
ceased and 5 days later the balancing tank between al., 1980). The acidogenic granules found in this
the reactors was removed. As can be seen from Fig. study were devoid of the multicellular filaments and
6 the alkalinity of the system steadily dropped from appeared to be bound together by chains of a large
7.0 g 1-l to less than 6.0 g 1-l over the following 10 rod-shaped bacterium with the surface of the granule
d. In order to maintain alkalinity within the system a predominantly consisting of short plump rods and
recycle of effluent from reactor M to reactor A was various coccus types. There appeared to be a smaller
installed with a recirculation ratio of 0.33. This was variety of morphological forms present within the
successful in preventing any further loss of alkalinity acidogenic granules than in the methanogenic ones,
which stabilized at around 5.3 g 1-l giving an and the density of packing of the bacteria within the
influent feed pH of 6.0 to reactor M. At an organic methanogenic granules appeared a lot higher; the
loading less than 28.2 g 1-l d- ’ COD the ratio of latter is likely to effect strength and settlement char-
volatile acid to alkalinity could be maintained in the acteristics (Prensier et al., 1988).
range 0.1-0.2 which is suitable for stable digestion Sludge granulation is a dynamic process without a
and methanogenesis. clear starting point and the factors which initiate this
process and influence granular formation are not
3.4. Sludge granulation fully understood (Dolfing et al., 1986; Colleran,
1988). One factor which is thought to be important is
Both reactors were seeded with non granular the superficial biogas developing rate and these were
sludge from single-stage digesters and as such were calculated at the time the granules were formed in
acclimated gradually at low loading rates to the new both of the reactors. The values found were 1.7 and
feedstock and reactor conditions. The first signs of 5.4 m dd ’ for the acidogenic and metbanogenic
granulation of the sludge were in the acidogenic reactors, respectively, and as such the value for the
reactor on day 80 when the influent SS concentration acidogenic reactor is less than previously reported
and the loading rate were 5.4 g 1-l and 16.6 g 1-l rates which lie in the range 3.6 to 10 (Wu et al.,
dd ’ COD, respectively. The granules were 2-3 mm 1985; Wiegant and Lettinga, 1985). This result was
in size and had an offensive odour. Granulation in not unexpected as overall the rate of gas production
the methanogenic reactor was found after 110 days in an acidogenic reactor is much lower than in a
of operation, when the loading rate was 22.1 g 1-l methanogenic reactor as the reaction products (VFA)
d- ’ COD, and the granules had an average diameter are liquid with only small quantities of H, and CO,
of l-2 mm. being evolved.
The granules that formed in the methanogenic and
acetogenic reactors differed in a number of ways
other than those mentioned above. The acidogenic 4. Conclusions
granules were much more fragile, appeared less
dense, and had a much lower settling velocity. Mor- The results clearly show that it is possible to split
phological differences were apparent on examination the anaerobic digestion process into two distinct
using SEM. The methanogenic granules were similar stages in a pair of upflow anaerobic sludge blanket
to those previously reported in UASB reactors treat- reactors. The acidogenic reactor, although tolerant to
ing food industry wastewaters (Wiegant and Let- an influent suspended solids concentration of 5.4 g
tinga, 1985; Dolfing et al., 1986; Dinopoulou and 1-l) showed signs of reduced solids liquification
Sterrit, 1988; Goodwin et al., 1992) in that they when the loading was increased and the suspended
consisted of networks of long multicellular filaments, solids concentration was 10.8 g l- ’ .
typical of Merhanothrix spp., with a diverse selec- Based on the efficiency indicators used it is clear
tion of rod and coccus type bacteria entrapped in a that the methanogenic reactor can quickly acclimate
dense matrix. The process of granulation under aci- to a feed which comprises the effluent from the
134 R. Borju et al. / Journd c$‘Biotechnology 45 (1996) 125-135

acidogenic reactor. Although loadings as high as 60 Chm, K.K. (1981) Anaerobic treatment kinetics of palm oil
sludge. Water Res. IS, 199-202.
g 1-l dd’ COD were applied these resulted in a
Chin, K.K., Ng, W.J., Ma, A.N. and Wong, K.K. (1987) Treata-
significant loss in COD removal efficiency and car- bihty studies of palm oil refinery wastewaters. Water Sci.
bon flow from longer chain fatty acids to methane. A Technol. 19, 23-29.
maximum load of around 30 g 1-l d-’ COD should Cohen, A. and Zoetemeyer, R.J. (1979) Anaerobic digestion of
ensure an overall 90% reduction in COD and effi- glucose with separated acid production and methane fortna-
tion. Water Res. 13, 571-580.
cient acid conversion to methane.
Colleran, E. (1988) Report on the technological aspects of granu-
Recirculation of effluent from the methanogenic lation. Proceedings of the GASMAT Workshop, Lunteren,
reactor to the acidogenic reactor proved to be an Netherlands, pp. 237-240.
effective way of maintaining reactor pH around 6 Davis, J.B. and Reilly, P.J.A. (1980) Palm oil mill effluent - A
which was suitable for efficient methanogenesis; this summary of treatment methods. Oleagineux 35, 323-30.
Dinopoulou, G. and Sterrit, R.M. (I 988) Anaerobic digestion of a
would save on the addition of chemicals and reduce
complex wastewater: kinetics of growth inhibition and product
overall running costs on a large-scale process and formation. Biotechnol. Bioeng. 3 I, 969-978.
also alleviate the need for an intermediate balancing Dolting, J., Griffioen, A., Van Neerven, A.R.W. and Zeven-
tank between the reactors. huizen, L.P.T.M. (1986) Chemical and biological compositiom
Granulation was observed in both reactors but of granular methanogenic sludge. Can. J. Microbial. 3 1, 744
750.
their morphological characteristics were distinct. The
Edewor. J.O. (1986) A comparison of treatment methods for palm
mechanism of granulation was not elucidated but oil mill effluent (POMEJ wastes. J. Chem. Technol. Biotech-
was shown to occur at typical values of superficial nol. 36, 212-218.
biogas developing rates in the methanogenic reactor Ghosh, S., Conrad, J.R. and Klass, D.L. (1975) Anaerobic acido-
and at values lower than previously reported in the genesis of wastewater sludge. J. Water Pollut. Control Fed. 47,
30-45.
acidogenic reactor.
Goodwin, J.A.S., Wase, D.A.J. and Forster, C.F. (1992) Pre-
granulated seeds for UASB reactors: how necessary are they.
Biores. Technol. 41, 71-79.
Acknowledgements Hanaki, K., Matsuo, T., Nagase, M. and Tabata, Y. (1987)
Evaluation of effectiveness of two-phase anaerobic digestion
process degrading complex substrate. Water Sci. Technol. 23,
The authors wish to thank the ‘Departamento de
1189-1200.
Postgrado y Especializacion de1 CSIC’ (Spain) for Ho, C.C. and Tan, Y.K. (1985) Anaerobic treatment of palm oil
financial support granted for the realization of this mill effluent by tank digesters. J. Chem. Technol. Biotechnol.
work. 35, 155-164.
Ho, C.C. and Chan, C.Y. (1986) The application of lead dioxide-
coated titanium anode in the electroflotation of palm oil mill
effluent. Water Res. 20, 1523- 1527.
References Hulshoff PO], L. and Lettinga. Cl. (1986) New technologies for
anaerobic wastewater treatment. Water Sci. Technol. 18, 41-
Abdul, K.. Mohamed, 1. and Kamil, A.Q.A. (1989) Biological 53.
treatment of palm oil mill effluent using Trichodermu uiridr. Hum, K.S., Thanh, N.C. and Lee, T.L. (1981) Palm oil waste
Biol. Wastes 27, 143-152. treatment study in Malaysia. Effluent Water Treat. J. 21,
American Public Health Association (APHA) (1985) Standard 452-455.
Methods for the Examination of Water and Wastewater, 16th Ibrahim, A., Yeoh. B.G., Cheah, SC., Ma, A.N., Ahmad. S.,
edn. APHA, Washington, DC. Chew, T.Y., Raj, R. and Wahid, M.J.A. (1984) Thermophilic
Badri, M.A. (1984) Identification of heavy metal toxicity levels in anaerobic contact digestion of palm oil mill effluent. Water
solid wastes by chemical specification. Conserv. Recycl. 7, Sci. Technol. 17, 155-166.
257-269. John, C.K. (1985) Treatment of agro-industrial wastes using water
Borja, R. and Banks, C.J. (1994) Anaerobic digestion of palm oil hyacinth. Water Sci. Technol. 17, 78 l-790.
mill effluent using an up-flow anaerobic sludge blanket reac- Karim, M.I.A. and Hie, L.L. (1987) The use of coagulating and
tor. Biomass Bioenergy 6, 381-389. polymeric flocculating agents in the treatment of palm oil mill
Cail, R.G. and Barford, J.P. (1985) Mesophilic semi-continuous effluent (POME). Biol. Wastes 22, 209-218.
anaerobic digestion of palm oil mill effluent. Biomass 7, Lettinga. G., Van Velsen, A.F.M., Hobma. S., De Zeeuw, W. and
287-295. Klapwijk, A. (1980) Use of the upflow sludge blanket reactor
R. Borju et al./Journal ofBiotechnology 45 (1996) 125-135 135

concept for biological wastewater treatment, especially for Prensier, G., Roustan, T.L., Dubourguier, H.C. and Albagnac, G.
anaerobic treatment. Biotechnol. Bioeng. 22, 699-734. (1988) Presence of viruses and other potential bacterial preda-
Lettinga, G. and Vinken, J.N. (1980) Feasibility of the upflow tors in granular methanogenic sludges. Proc. of the GAST-
anaerobic sludge blanket (UASB) process for the treatment of MAT Workshop, Lunteren, Netherland, pp. 96- 101.
low-strength waste. Proc. of the 35th Purdue Industrial Roge, W. and Velayuthan, A. (1981) Preliminary trials with
Wastewater Conference, pp. 635-644. Westfalia-3-phase decanters for palm oil separation. In: Push-
Ma, A.N. and Ong, ASH. (1986) Palm oil processing - new parajah, E. and Rajadurai, M. (Eds.), Palm Oil Prod. Technol.
development in effluent treatment. Water Sci. Technol. 18, Eighties, Rep. Proc. Int. Conf., Inc. Sot. Plant. Kuala Lumpur,
35-40. Malaysia. pp. 327-334.
Ma, A.N. and Ong, A.S.H. (1988) Treatment of palm oil steriliser Souza, M.E. (1986) Criteria for the utilization, design and opera-
condensate by an anaerobic process. Biol. Wastes 23, 85-97. tion of UASB reactors. Water Sci. Technol. 18, 55-69.
Mossey, M.L. and Pohland. F.G. (1978) Phase separation of Sutanto, J. (1983) Solvent extraction process to achieve zero-ef-
anaerobic stabilization by kinetic control. J. Water Pollut. fluent and to produce quality animal feed from mill sludge.
Control Fed. 50, 2204-2222. Planter 59, 17-35.
Ng, W.J., Wong, K.K. and Chin, K.K. (1985) Two-phase anaero- Suwandi, M.S. (1981) Retention characteristics of polyamide and
bic treatment kinetics of palm oil wastes. Water Res. 19, polysulfone membranes in relation to palm oil mill effluent.
667-669. Sains Malays. IO, 147-160.
Ng, W.J., Goh, A.C.C. and Tay, J.H. (1987a) Palm oil mill Turisin, M., Nor, M. and Suwandi, MS. (1981) Membrane pro-
effluent (POME) treatment - an assesment of coagulants used cesses in by-product recovery. Sains Malays. 10, 161- 174.
to aid liquid-solid separation. Biol. Wastes 21, 237-248. Vigneswaran, S., Balasuriya, B.L.N. and Viraraghavan, T. (1986)
Ng, W.J., Chin, K.K. and Wong, K.K. (1987b) Energy yields Anaerobic wastewater treatment-attached growth and sludge
from anaerobic digestion of palm oil mill effluent. Biol. blanket process. Environmental Sanitation Information Center,
Wastes 19, 257-266. Bangkok, Thailand.
Ng, W.J., Goh, A.C.C. and Tay. J.H. (1988) Palm oil mill effluent Wiegant, W.M. and Lettinga, G. (1985) Thermophilic anaerobic
treatment - liquid-solid separation with dissolved air flota- digestions of sugars in upflow anaerobic sludge blanket reac-
tion. Biol. Wastes 25, 257-268. tors. Biotech. Bioeng. 27, 1603-1607.
Pandey, A. and Ramakrishna, S.V. (1990) Management of efflu- Wu, W.M., Hu, J.C. and Gu, X.S. (1985) Properties of granular
ent from edible oil (palm oil) industry. In: Pollution Manage- sludge in upflow anaerobic sludge blanket (UASB) reactors
ment in Food Industries Symposium, pp. 159- 17 1. and its formation. Anaerobic Digestion 1985, Guangzhou,
Petitpierre, G. (1982) Le traitement des effluents d’huilerie de China, pp. 339-350.
palme et la production de biogaz. Oleagineux 37. 367-374. Zoetemeyer, R.J., Van den Heuvel, J.C. and Cohen, A. (1982) pH
Peyton, T.O., Cooper, I.W. and Quab, S.K. (1979) Anaerobic influence on acidogenic dissimilation of glucose in an anaero-
digestion of palm oil mill effluent. Proc. 34th Indust. Waste bic digester. Water Res. 16, 303-311.
Conf., Purdue University, Lafayette.

Das könnte Ihnen auch gefallen