Seediscussions,stats,andauthorprofilesforthispublicationat:https://www.researchgate.net/publication/7827473

Oliveoilmilleffluentsasafeedstockforproductionofbiodegradablepolymers.WaterRes39:2076-2084

ARTICLEinWATERRESEARCH·JUNE2005

ImpactFactor:5.53·DOI:10.1016/j.watres.2005.03.011·Source:PubMed

CITATIONS

118

READS

75

6AUTHORS,INCLUDING:

DavideDionisi

UniversityofAberdeen

34PUBLICATIONS999CITATIONS

SEEPROFILE

MarcoPetrangeliPapini

SapienzaUniversityofRome

60PUBLICATIONS1,131CITATIONS

SEEPROFILE

FulvioCarrasco

EscuelaSuperiorPolitécnicadelLitora…

123PUBLICATIONS1,952CITATIONS

SEEPROFILE

Allin-textreferencesunderlinedinbluearelinkedtopublicationsonResearchGate,

lettingyouaccessandreadthemimmediately.

Availablefrom:MarcoPetrangeliPapini

Retrievedon:04February2016

ARTICLE IN PRESS

0043-1354/$ - se

doi:10.1016/j.w

�CorrespondE-mail addr

Water Research 39 (2005) 2076–2084

www.elsevier.com/locate/watres

Olive oil mill effluents as a feedstock for production ofbiodegradable polymers

D. Dionisia,�, G. Caruccia, M. Petrangeli Papinia, C. Riccardib, M. Majonea,F. Carrascoc

aDepartment of Chemistry, University of Rome ‘‘La Sapienza’’, Piazzale Aldo Moro, 5, Rome 00185, ItalybNational Institute for Safety and Health (ISPESL), Via Urbana, 167, Rome 00184, Italy

cDepartment of Chemical Engineering, University of Girona, Campus Montilivi, Girona 17071, Spain

Received 3 November 2004; received in revised form 24 February 2005

Abstract

The aim of the present paper was to study the feasibility of using olive oil mill effluents (OMEs) as a substrate in

biodegradable polymer production. OMEs were anaerobically fermented to obtain volatile fatty acids (VFAs), which

are the most highly used substrate for polyhydroxyalkanotes (PHAs) production. The anaerobic fermentation step was

studied both without pretreatment and with different pretreatments (i.e., centrifugation, bentonite addition, and

bentonite addition followed by centrifugation) and at various concentrations (28.5, 36.7 and 70.4 gCODL�1). During

fermentation, VFA concentration was determined (7–16 gCODL�1) as well as the corresponding yield with respect to

initial COD (22–44%). At all initial concentrations, centrifugation pretreatment (with or without previous addition of

bentonite) significantly increased the final VFA concentration and yield, whereas the addition of bentonite alone had no

influence. Moreover, centrifugation pretreatment led to a different acid distribution, which affected the hydroxyvalerate

(HV) content within the obtained copolymer poly b-(hydroxybutyrate-hydroxyvalerate) [P(HB-HV)]. OMEs were

tested for PHA production by using a mixed culture from an aerobic SBR. Centrifuged OMEs, both with or without

fermentation, were tested. PHAs were produced from both matrices, but with fermented OMEs PHA production was

much higher, because of the higher VFA concentration. The initial specific rate of PHA production obtained with

fermented OMEs was approximately 420mg CODgCOD�1 h�1 and the maximum HV content within the copolymer

was about 11% (on a molar basis). The HV monomer was produced only until propionic acid remained present in the

medium.

r 2005 Elsevier Ltd. All rights reserved.

Keywords: Oil mill effluents; Volatile fatty acids; Polyhydroxyalkanoates; Pretreatments; Acidogenic fermentation; Polymer storage

1. Introduction

Bioplastics are attracting much interest as alternatives

to traditional plastics, this is due to the fact that they are

e front matter r 2005 Elsevier Ltd. All rights reserve

atres.2005.03.011

ing author. Fax: +3906490631.

ess: davide.dionisi@uniroma1.it (D. Dionisi).

biodegradable and can be formed from renewable

resources. Among the biodegradable plastics polyhy-

droxyalkanoates (PHAs), and in particular the copoly-

mer poly b-(hydroxybutyrate-hydroxyvalerate) [P(HB-

HV)], are the most promising. P(HB-HV) has similar

properties to polypropylene, it can be processed in the

same way and it could have the same wide range of

d.

ARTICLE IN PRESSD. Dionisi et al. / Water Research 39 (2005) 2076–2084 2077

application (Lee, 1996; Carrasco et al., 2004). However,

until now production of P(HB-HV) has proven to be

much more costly than traditional oil-derived plastics

and this has hindered its use becoming more widespread

(Salehizadeh and van Loosdrecht, 2004). Hence, there is

a potential for widening the market for PHAs, provided

that their cost decreases.

One of the most relevant costs in the production of

PHAs is the cost of the substrate (Lee, 1996; Choi and

Lee, 2000), which is usually pure (e.g. glucose and

propionate for P(HB-HV) production). In recent years,

the use of organic wastes (such as swine waste liquor,

palm oil mill effluents, and vegetable and fruit wastes) is

being studied as an alternative substrate for PHA

production (Hassan et al., 1996, 1997a, b; Meesters,

1998; Reis et al., 2003; Salehizadeh and van Loosdrecht,

2004).

Olive oil mill effluents (OMEs) can also be considered

as an alternative, no-cost substrate for PHA production.

OMEs cause serious environmental pollution due to

their high level of production (about 30 millionm3/year

in Mediterranean countries) and COD content

(100–200 gCODL�1). Up to now, the processes used

to treat OMEs are considerably diversified: combustion

fluidized bed (Cliffe and Patumsawad, 2001), anaerobic

filter (Borja and Banks, 1995), anaerobic digestion and

codigestion (Andreozzi et al., 1998; Angelidaki and

Ahring, 1997), UASB (Saatci et al., 2003), membrane

technology (Ahmad et al., 2003), pond system (Chin et

al., 1996), ozonation reactor (Andreozzi et al., 1998),

and activated carbon bed (Galiatsatou et al., 2002).

From the point-of-view of PHA production, their

high COD content can be an advantage due to the

possibility of fermenting OMEs at high rates to produce

volatile fatty acids (VFAs) (Beccari et al., 1996), which

in turn can be used for PHA production. Indeed, the

acidogenic fermentation step is not inhibited to a great

extent by OME lipidic and phenolic substances, which,

on the other hand, inhibit the further step of methano-

genesis (Beccari et al., 1999). Another advantage of

OMEs with respect to other organic wastes is that they

are a liquid stream and can therefore be easily pumped

and treated in slurry-phase bioreactors. There is a lack

of experimental data on the feasibility of using

fermented OMEs for PHA production, however.

A new process (Dionisi et al., 2004) was proposed for

the production of biodegradable PHAs from wastes, by

using a combination of anaerobic and aerobic steps. The

proposed process has three sequential steps. In the first

step, acidogenic fermentation transforms highly concen-

trated biodegradable waste into a mixture of organic

acids. In the second step, an aerobically activated sludge

process is operated at medium–high organic load by

periodic feeding in a Sequencing Batch Reactor (SBR) in

order to enrich and produce a sludge with a high storage

response. The produced excess sludge has a high storage

response that is exploited in a third step, this is operated

in batch but at a considerably higher organic load in

order to saturate the sludge storage capacity. The PHA-

rich sludge then flows downstream for processing for

PHA extraction and purification. This process has only

been previously studied with reference to steps 2 and 3,

however, with only synthetic mixtures of organic acids

being used as the feed (Dionisi et al., 2004). Experi-

mental studies on applications with real waste matrices

are therefore lacking.

The aim of this study was at to examine the

production of PHAs from OMEs, with both the

fermentation step (step 1) and the PHA production step

(step 3) being investigated. In the fermentation step,

VFA production rates and yields were studied for OMEs

with and without different pretreatments (bentonite

addition, centrifugation, or bentonite addition followed

by centrifugation). The PHA production stage was

investigated by means of batch tests performed by using

a mixed culture enriched in a SBR fed with a mixture of

organic acids (step 2, Dionisi et al., 2004). The

production rates and the composition of the biodegrad-

able polymer were also investigated in comparison with

synthetic substrates to which the enriched biomass had

been acclimated.

2. Experimental methods

2.1. OME characterization

Fresh OMEs were obtained from an olive continuous

centrifuge processing plant. The OMEs were character-

ized as follows: pH ¼ 5.2, total COD ¼ 113.8 gL�1,

soluble COD (filtered at 0.22 mm) ¼ 34.5 gL�1, soluble

polyphenols¼ 2.2gL�1, soluble carbohydrates ¼ 3.1gL�1,

Kjeldhal nitrogen ¼ 2gL�1 and lipids ¼ 11.1gL�1.

2.2. Pretreatment tests

The OMEs were brought to pH 6.5 by adding CaO.

They were then used directly for fermentation tests (see

Section 2.3) or pretreated according to three different

procedures. The first type of pretreatment was carried

out by the addition of bentonite (B-3378 Sigma) (up to

10 gL–1, Beccari et al., 2001). The bentonite–OME

slurry was magnetically stirred for 2 h, then it was left

for 24 h. The bentonite–OME slurry was either recov-

ered and used in its entirety for fermentation tests

(OME–B) or it was pretreated further by centrifuging

(5000 rpm for 15min), and the supernatant was used in

the fermentation tests (second type of pretreatment,

OME–BC). The third type of pretreatment was the

direct centrifugation of OME with no previous addition

of bentonite (OME–C). The supernatant was recovered

and used for fementation tests, also in this case.

ARTICLE IN PRESSD. Dionisi et al. / Water Research 39 (2005) 2076–20842078

2.3. Biotreatability tests

Batch tests were carried out in 500mL glass bottles

with perforated screw caps and latex under-caps. These

bottles were filled with 200mL of slurry with the

following composition: OMEs or pretreated OMEs

(50–130mL), inoculum (10–20mL, up to 10% of total

COD) and distilled water (as required for a final volume

of 200mL). Within the given ranges, the added volume

of OMEs or pretreated OMEs varied for the different

series of fermentation tests (see below) and for the

different pretreatments, being adjusted in order to reach

the desired COD at the beginning of the test. In order to

produce and maintain anaerobic conditions, each bottle

was fluxed with CO2–N2 (at the beginning of the test as

well as after each sampling). Inoculum was obtained

from an anaerobic digester treating lipid-rich food

factory effluents. Bottles were kept at 25 1C under

continuous agitation at 150 rpm in a thermostatic bath

(Dubnoff BSD/D). For all OME samples, the initial pH

was brought to 6.5 by the addition of CaO. Three

different series (I–III) of fermentation tests were

performed at various initial concentrations (28.5, 70.4

and 36.7 gCODL�1, respectively). The highest concen-

tration (70.4 gCODL�1) corresponded to the lowest

OME dilution that could be successfully treated under

methanogenic conditions (Beccari et al., 2001). All

bottles had the same initial COD within each series.

2.4. Polymer production tests

Two different OME matrices were used for PHA

production tests: (1) centrifuged and fermented OME

and (2) only centrifuged OME (without fermentation).

The biomass for PHA production tests was an activated

sludge enriched under ‘‘feast and famine’’ aerobic

conditions in a SBR. The operating conditions of the

SBR were the same as in a previous study (Dionisi et al.,

2004): the feed was a mixture of acetic, lactic and

propionic acids (40%, 40% and 20%, respectively, on a

COD basis) and the organic load rate was

8.5 gCODL�1 d�1. In perfoming the PHA production

tests, the biomass was withdrawn from the SBR at the

end of the cycle, and diluted to the chosen concentration

(about 300mgVSSL�1). A fixed volume (70mL) of the

chosen OMEs was then added to the biomass. The tests

were carried out at a pH of 7.5 and a temperature of

25 1C. Samples were taken at regular intervals during the

test in order to determine the VFA and PHA contents.

The oxygen uptake rate (OUR) was evaluated by

interrupting aeration for short periods and measuring

oxygen concentration as a function of time. Initial rates

of VFA removal and PHA formation were calculated by

linear regression of concentration vs time data and were

expressed as specific rates with respect to the initial

biomass concentration. The biomass concentration was

calculated as gCODL�1 by using a conversion factor of

1.42 gCODgVSS�1. Refer to Dionisi et al. (2004) for

further details regarding the performance of the test.

2.5. Analyses

In order to determine polyphenols in the liquid phase

the Folin Ciocalteau method was used. Lipids were

evaluated by gravimetric analysis. Carbohydrates were

analyzed by spectrophotometry upon color development

by reaction with phenol and sulfuric acid. VFA and

lactic acid were determined by GC (Perkin Elmer,

column: packed stationary phase Carbograph 1 AL 80/

120, L ¼ 2:4m, D ¼ 3:2mm, carrier N2, detector FID,

internal standard: acrylic acid). PHB and P(HB-HV)

were hydrolyzed and esterified to 3-hydroxyacyl methyl

esters and then determined by GC (Perkin Elmer,

stationary phase 2% Reoplex 400 on 80/120 Chromo-

sorb GAW, L ¼ 1:83m, D ¼ 2mm, gas carrier N2

detector FID, internal standard: benzoic acid). HB and

HV monomers were converted into COD units accord-

ing to oxidation stoichiometry: 1.38mgCODmg(HB)�1

and 1.63mgCODmg(HV)�1.

In biotreatability tests, the gaseous phase was

analyzed for methane and hydrogen content by GC

(Varian 3400, 2mm packed column, stationary pha-

se ¼ Carbopack B 1%, gas carrier He, FID, injector and

detector T ¼ 260 1C, isotherm mode ¼ 80 1C). More

detailed descriptions of the analytical methods are

reported elsewhere (Beccari et al., 1999; Dionisi et al.,

2004). COD, N-NH3 and solid measurements were made

according to APHA–AWWA–WPCF procedures

(1992).

3. Results and discussion

3.1. Pretreatment effects on COD removal from OME

Total COD removal (mean value of three replicates)

was determined for each pretreatment. Centrifugation

(C) and addition of bentonite followed by centrifugation

(BC) showed similar COD removal efficiency: 55.8%

and 57.7%, respectively (Table 1). When bentonite was

added without subsequent separation (B) there was no

COD removal as it remained in the slurry and, even if

partially adsorbed on bentonite, was potentially avail-

able for the subsequent fermentation step.

3.2. Pretreatment effect on rate and yield of VFA

production

Two series of fermentation tests were performed

simultaneously to evaluate the effect of bentonite at

initial concentrations of 28.5 gCODL�1 (Series I) and

70.4 gCODL�1 (Series II), respectively. At both initial

ARTICLE IN PRESS

18.0

16.0

14.012.0

10.0

8.0

6.0

4.0

2.0

0.00 5 10 15 20 25 30 35 40 45 50

Time (days)V

FA p

rodu

ctio

n, g

CO

D/L

Series I, OMESeries I, OME-BSeries I, OME-BCSeries II, OMESeries II, OME-B

Fig. 1. Comparison of fermentation tests, by using effluents

without pretreatment (OME) or pretreated with bentonite alone

(OME–B) or bentonite+centrifugation (OME–BC). Initial

OME concentration: 28.5 gCODL�1 (Series I) and

70.4 gCODL�1 (Series II).

Table 1

COD removal for the different pretreatments

Bentonite addition

OME–B

Bentonite addition+centrifugation

OME–BC

Centrifugation

OME–C

Final COD content (gCODL�1) 113.8 48.1 50.3

COD removal (%) 0 57.7 55.8

D. Dionisi et al. / Water Research 39 (2005) 2076–2084 2079

concentrations, tests were performed with (OME–B) and

without (OME) bentonite addition. Tests with bentonite

addition (OME–B) aimed at verifying whether its

addition and maintenance during the test could improve

acid production with respect to OME that were not

pretreated. A similar possible positive effect has been

already shown for methanogenesis (Beccari et al., 2001),

due to initial adsorption of inhibiting compounds (thus

increasing the initial conversion rate) and their slow

release as the fermentation then continues (thus main-

taining the whole COD available for conversion).

In order to separate adsorption and release effects,

fermentation tests with bentonite addition and then

separation of adsorbed COD by centrifugation before

fermentation were also performed (OME–BC). It was

not possible to perform this type of test at the highest

organic load (70.4 gCODL�1), however, due to the loss

of COD due to centrifugation (see Table 1).

As shown in Fig. 1, VFAs were produced in all tests

during the first week, then acid production continued at

a considerably slower rate and a plateau was reached at

20–25 days. In all tests, the pH dropped to 5 within the

first week and remained stable. Hence, on day 28, CaO

up to pH 6 was added but little or no effect was observed

and a new plateau was reached at 40–50 days. Gas

analysis during the whole test demonstrated a low

production of methane and hydrogen, lower than 3%

and 1% of the total COD in all tests, respectively. The

final VFA concentrations and yields are reported in

Table 2.

With OMEs that were not pretreated, at the highest

initial concentration, VFA production was faster (Fig. 1)

and reached a higher final concentration (Table 2)

(15.6 vs. 7.1 gCODL�1 for 70.4 and 28.5 gCODL�1,

respectively). As a result, the final VFA yield at

70.4 gCODL�1 was only slightly lower than that at

28.5 gCODL�1 (22.1% vs. 24.9%). This indicates that,

in this range of concentrations, COD exerts no or low

inhibiting effects on acidogenesis. Accordingly, the

presence of bentonite during fermentation did not exert

a positive effect on the initial rate or final yield. These

results are different from those previously reported for

methanogenesis (Beccari et al., 2002), thus confirming

that acidogenesis is less sensitive than methanogenesis to

the inhibition effects of lipids and polyphenols. On the

other hand, conversions are lower than those usually

reported for methanogenesis under non-inhibiting con-

ditions, within the range of 70–80% (Beccari et al.,

1996). This is probably due to the fact that a fraction of

the COD in OME and OME–B is of a lipidic and

phenolic nature and requires syntrophic methanogenic

activity in order to be degraded (i.e. H2 scavenging

through interspecies transfer).

A better performance was observed with the addition

of bentonite and further centrifugal separation

(OME–BC) was performed before the fermentation

tests. At an initial concentration of 28.5 gCODL�1,

OME–BC showed both a faster VFA production (Fig.

1) and a higher yield (Table 2) than OME or OME–B. In

particular, the final yield was 70–75% higher (44% for

OME–BC vs. 25% for OME and 26% for OME–B).

Given that the initial COD concentration was exactly

the same in all tests, observed changes must be

attributed to qualitative changes of the COD composi-

tion due to centrifugation. As higher VFA production

yields were achieved, it follows that the residual COD

left by centrifugal separation was more fermentable or,

in other words, the separation of the bentonite–OME

slurry succeeded in selectively removing the less fermen-

table COD fraction, such as lipids and polyphenols. The

higher VFA yield of OME–BC allowed a final VFA

concentration that was only 20% lower (12.4 vs.

15.6 gCODL�1) to be reached, compared with a 2.5

times higher initial concentration with no pretreatment.

This confirms that the loss of COD due to centrifugation

of bentonite-OME slurry is not so relevant in terms of

ARTICLE IN PRESS

Table 2

VFA production and distribution of acids, at different initial concentrations and for various pretreatments, at the end of fermentation

(OME: without pretreatment, OME–B: bentonite addition, OME–C: centrifugation, OME–BC: bentonite addition+centrifugation)

Series I Series II Series III

OME OME–B OME–BC OME OME–B OME OME–BC OME–C

Initial OME (gCODL�1) 28.5 28.5 28.5 70.4 70.4 36.7 36.7 36.7

Produced VFA (gCODL�1) 7.1 7.3 12.4 15.6 15.5 9.2 12.0 13.0

VFA yield (%) 24.9 25.9 43.6 22.1 22.0 25.2 32.5 35.6

% of each acid within VFA

Acetic acid 53.4 52.7 54.4 54.1 51.7 46.7 31.6 38.7

Propionic acid 14.7 14.0 7.1 9.8 9.0 16.4 9.5 9.9

Isobutyric acid 0.4 0.1 2.5 0.3 0.1 0.0 2.0 2.7

Butyric acid 27.4 29.5 39.5 37.3 36.3 21.8 54.4 46.7

Valeric acid 4.1 3.7 0.5 3.4 2.9 15.1 2.5 2.1

14.0

12.0

10.0

8.0

6.0

4.0

2.0

0.00 5 10 15 20 25 30 35 40

Time (days)

VFA

pro

duct

ion,

g C

OD

/L

Series III, OME

Series III, OME-C

Series III, OME-BC

Fig. 2. Comparison of fermentation tests by using effluents

without pretreatment (OME) or pretreated with centrifugation

alone (OME–C) or bentonite+centrifugation (OME–BC).

Initial OME concentration: 36.7 gCODL�1 (Series III).

D. Dionisi et al. / Water Research 39 (2005) 2076–20842080

VFA production because of its low biodegradability

under acidogenic conditions.

Based on these results, a third series of fermentation

tests (Series III) was performed to ascertain whether

such a positive effect on VFA production rate and yield

was actually due to the previous separation of the

bentonite-OME slurry (where some COD is adsorbed on

the bentonite) or only due to the centrifugation of

OMEs. OMEs with no pretreatment (OME) were

therefore compared with two different centrifugation

pretreatments: with (OME–BC) or without (OME–C)

previous bentonite addition. The initial concentration

was taken at the highest value possible (36.7 gCODL�1)

considering the COD removal due to centrifugation (see

Table 1) and dilution due to the addition of inoculum to

start the fermentation test. As in Series I and II, early

VFA production was observed in all samples for the first

few days of fermentation, then the VFA production rate

decreased and a plateau was reached after about 25 days

(see Fig. 2). Both the initial VFA production rate and

the final VFA concentration were lower for OMEs

without pretreatment than those for pretreated

OME–BC and OME–C. Moreover, the VFA concentra-

tion profiles were quite similar for OME–BC and

OME–C, thus confirming that centrifugation was

effective by itself, independent of whether the bentonite

was added previously or not. As reported in Table 2,

centrifugation pretreatment caused a 30–40% relative

increase in VFA yields as compared to non-pretreated

OME.

By comparing the three series as a whole (Table 2),

centrifugation and fermentation at a medium initial

concentration (36.7 gCODL�1, Series III) allows a VFA

concentration to be reached that is only slightly lower

(13.0 vs. 15.6 gCODL�1) than that obtained with no

centrifugation and fermentation at an almost double

initial concentration (70.4 gCODL�1, Series II). Thus,

even though pretreatment by centrifugation limits the

maximum initial COD concentration in the fermenta-

tion step, this pretreatment is the best compromise in

relation to the subsequent polymer production stage,

because high VFA yield and concentration were

reached. Indeed, even if fermentation of non-centrifuged

OMEs at a high initial concentration produces a slightly

higher VFA concentration (15.6 gCODL�1), yields were

far lower (22.1%). This means that a larger quantity of

non-acidified COD remains that will eventually consume

large oxygen quantities in the polymer production stage

(with a consequent cost) and the associated biomass

growth will lead to a significant decrease of the polymer

production yield and of polymer/biomass fraction. It

must be noted that the best VFA concentration at the

intermediate concentration of 36.7 gCODL�1

(13 gCODL�1) was higher than the best reached

(9 gCODL�1) by using palm oil mill effluents, as

utilized in a previous study (Hassan et al., 1996).

ARTICLE IN PRESS

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0-50 0 50 100 150 200 250 300 350 400

Time (min)

But

yric

and

ace

tic a

cid

(g C

OD

L-1

)

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

Pro

pion

ic a

nd is

obut

yric

aci

d(g

CO

D L

-1)

acetic acidbutyric acidpropionic acidisobutyric acid

0

100

200

300

400

500

600

-50 0 50 100 150 200 250 300 350 400Time (min)

HB

(m

g C

OD

L-1

)

0

5

10

15

20

25

30

35

40

45

HV

(m

g C

OD

L-1

)

HBHV

0

10

20

30

40

50

60

-50 0 50 100 150 200 250 300 350 400Time (min)

OU

R (

mgO

2 L-1

h-1

)

(A)

(B)

(C)

Fig. 3. PHA production tests using fermented OME–C (with

centrifugation pretreatment).

D. Dionisi et al. / Water Research 39 (2005) 2076–2084 2081

3.3. Pretreatment effects on acids distribution within

VFAs

Table 2 also shows the relative distribution of single

acids within VFAs (as a percentage of the total VFAs in

terms of COD) at the end of fermentation. With regard

to OMEs with no pretreatment, acetic acid was the main

product in the whole range of initial concentrations

(47–54%), with lower relative amounts of butyric acid

(22–37%), propionic acid (10–16%) and valeric acid

(4–15%). Isobutyric and lactic acids were produced in

negligible quantities or not produced at all. Tests where

OMEs were only pretreated by bentonite addition

(OME–B) showed a very similar distribution, thus

confirming that the addition of bentonite did not

significantly change the spectrum of compounds that

were made available to acidogenic fermentation.

The relative distribution of single acids changed

completely when OMEs were centrifuged (OME–C or

OME–BC). In this case, the relative abundance of

butyric acid increased with respect to untreated OMEs

in Series III: 47–54% vs. 22%, whereas the opposite was

found for acetic acid (32–39% vs. 47%), propionic acid

(10% vs. 16%) and valeric acid (2% vs. 15%). This

change of relative distribution of VFAs confirms that

centrifugation greatly changes the spectrum of com-

pounds that are available to acidogenic fermentation.

The different VFA distribution can have a great

influence on the composition of the polymer produced in

the subsequent step, namely on the percentage of HV

monomers within the copolymer P(HB-HV). Indeed,

VFAs containing an even number of carbon atoms (i.e.

acetic and butyric acids) mostly lead to the formation of

HB monomers, whereas VFAs containing an odd

number of carbon atoms (i.e. propionic and valeric

acids) mostly lead to the formation of HV monomers

(Yu, 2001; Dionisi et al., 2004). As an increase of HV

content within the copolymer generally improves its

thermal and mechanical properties, the negative effect of

centrifugation on propionic and valeric acid production

has to be considered along with the important beneficial

effects discussed above.

3.4. PHA production tests

PHA production experiments were run with both

centrifuged and fermented OME and with centrifuged

OME that had not been fermented. A third type of PHA

production tests was also performed by using the

synthetic medium to which the biomass was previously

acclimated (see experimentals).

Fig. 3 shows a PHA production test in which

centrifuged and fermented OME were added to the

enriched biomass. Most VFAs were removed simulta-

neously (acetic, propionic and butyric acids), the only

exception being isobutyric acid. As seen in Fig. 3a, the

highest initial specific removal rate was that of butyric

acid (269mgCODgCOD�1 h�1, with respect to the

initial biomass concentration) whereas acetic and

propionic acids were removed at considerably lower

rate (95 and 91mgCODgCOD�1 h�1, respectively). The

higher removal rate of butyric acid with respect to acetic

and propionic acid was unexpected, since the biomass

had not been previously acclimated to butyric acid

whereas it had been acclimated to acetic and propionic

acids. As a whole, the initial specific removal rate of all

VFAs was 455mgCODgCOD�1 h�1 (6% standard

deviation on three replicates).

The initial specific storage rate of PHAs (Fig. 3b) was

298mgPHAgCOD�1 h�1 (417mgCODgCOD�1 h�1,

expressing PHA as equivalent COD) (10% standard

deviation). The formation of the HV monomer within

ARTICLE IN PRESS

0

500

400

300

200

100

600

0 50 100 150 200 250 300 350 400

Time (min)

0 50 100 150 200 250 300 350 400

Time (h)

Tot

al P

HA

s (

mgC

OD

L-1

)

Synthetic substrates

Fermented OMEs

0

500

3000

2500

2000

1500

1000R

emov

ed o

rgan

ic a

cids

(VF

As

+ L

actic

aci

d) (

mgC

OD

L-1

)

Synthetic substrates

Fermented OMEs

(A)

(B)

Fig. 5. Comparison of tests with fermented OME–C (shown in

Fig. 3) and those with synthetic substrates: (A) stored polymers

and (B) removed organic acids (VFAs and lactic acid).

D. Dionisi et al. / Water Research 39 (2005) 2076–20842082

the copolymer P(HB-HV) occurred for the period of

time in which propionic acid was present in the medium

(about 200min). The maximum HV content was 11%,

on a molar basis. At the end of the test the polymer

concentration was approximately 350mgPHAL�1 and

the corresponding polymer content in the biomass was

0.54mgPHAmgVSS�1 (calculated assuming that no

growth occurs during the test, as indicated by the OUR

profile). The final HV content in the polymer was 4%

(on a molar basis).

A comparison of the profiles of VFAs (Fig. 3a), PHAs

(Fig. 3b) and OUR (Fig. 3c) shows that these profiles are

not perfectly linked to each other. In the last part of the

experiments, both the PHA formation rate and OUR

showed a decreasing trend that was not present for the

overall VFA removal. This behavior likely indicates that

readily biodegradable substrates other than acids were

initially contributing to PHA formation and oxygen

consumption. This is also confirmed by comparing

overall VFA consumption and PHA formation through-

out the experiment. Fig. 4 shows that the two profiles

were almost corresponding for about 4 h, which would

correspond to a yield of about 1mgPHAmgVFA�1

(both expressed as COD). This storage yield is

practically impossible because storage requires energy

and a part of the substrate has to be oxidized to support

it. The thermodynamic storage yield from acetate has

been reported to be 0.73mgPHAmgAcid�1 (expressed

as COD) (van Aalst-van Leeuwen et al., 1997). In

another previous study (Hassan et al., 1997a), where

palm oil mill effluents and a pure culture were used, the

yield was significantly lower: 0.22mgPHAmgVFA�1

(or 0.30, expressing PHA and VFA as COD).

Fig. 5 shows the comparison of PHA production from

centrifuged and fermented OME and from the synthetic

medium containing only organic acids (VFAs and lactic

acid). PHAs were formed at similar rates from both

media, thus confirming that the biomass was as capable

of using the new feed as the synthetic feed to which it

had been previously acclimated. This also indicated that

0

100

200

300

400

500

600

700

800

900

1000

0 50 100 150 200 250 300 350 400Time (min)

Rem

oved

VF

As

and

stor

ed P

HA

s(m

g C

OD

L-1

)

Removed VFAs

Stored PHAs

Fig. 4. Comparison between removed VFA and stored PHA in

the test shown in Fig. 3.

centrifuged and fermented OMEs did not contain

inhibiting compounds (at least at the chosen concentra-

tion). On the other hand, it is evident that the removal of

organic acids was much higher in the synthetic feed than

in the OME–C, confirming that in the latter case carbon

sources other than VFA directly contribute to the

metabolism of PHA formation. It must be noted that

conversion of organic acids to PHAs in the synthetic

medium was 0.39mgPHAmgVFA�1 (expressed as

COD), similar to values previously reported (Dionisi

et al., 2004).

Fig. 6 shows the results of a typical test in which

centrifuged and non-fermented OMEs were used for

PHA production. The initial concentration of organic

acids was much lower than that obtained by using

fermented OME–C, as previously reported. For this

reason, the non-fermented OME–C mixture was re-

added twice during the test. Moreover, the non-

fermented OME–C contained only acetic and lactic

acids. Fig. 6 shows that both acids were removed

simultaneously, the lactic acid being quickly depleted

(initial rate: 770mgCODgCOD�1 h�1). The homopo-

lymer PHB was formed instead of the copolymer

P(HB–HV), produced from fermented OME–C. As a

result of the very high rate of lactic acid removal, the

initial PHA formation rate was also high

ARTICLE IN PRESS

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

00 50 100 150 200 250 300 350 400

Time (min)

Ace

tic a

nd la

ctic

aci

ds (

mgC

OD

L-1

)

Acetic acid

Lactic acid

OME re-addition OME re-addition

0

20

40

60

80

100

120

-50 0 50 100 150 200 250 300 350 400

Time (min)

HB

and

HV

(m

gCO

D L

-1) HB

HV

(A)

(B)

Fig. 6. Batch test with non fermented OME–C (with centrifu-

gation pretreatment).

0

600

500

400

300

200

100

0 50 100 150 200 250 300 350 400

Time (min)

Tot

al P

HA

s (m

gCO

D L

-1)

non fermented OMEsfermented OMEs

Fig. 7. Comparison of the produced PHA in the tests with

fermented and non-fermented OME–C. (NB: For a correct

comparison, PHA data of the test with non-fermented OMEs

have been corrected taking dilution due to OME re-addition

into account.)

D. Dionisi et al. / Water Research 39 (2005) 2076–2084 2083

(523mgCODgCOD�1 h�1) but quickly decreased after

lactic acid depletion. Moreover, in this case the yield of

PHB on removed acids was approximately

0.40mgPHAmgVFA�1 (expressed as COD). Thus,

there was no evidence that substrates other than organic

acids could contribute to the metabolism of PHA

formation. This can be explained in terms of previous

fermentation having produced other substrates at low

molecular weights, that are more easily biodegradable

than parent compounds at higher molecular weights, as

were originally contained in centrifuged OMEs. After

the re-addition of non-fermented OME–C, a general

decrease of both acid removal and PHA formation rate

were observed, which could indicate a certain degree of

inhibition. As a consequence of several effects, Fig. 7

shows that the PHA formation by using fermented

OME–C was much higher than by using OME–C with

no previous fermentation: the PHA produced at the end

of the experiment was 465 vs. 152 gCODL�1, i.e. 3

times higher.

4. Conclusions

The experimental results showed that olive oil mill

effluents represent a potential feedstock for production

of PHAs from high storage capacity mixed cultures,

enriched under feast and famine conditions.

OME centrifugation plus fermentation demonstrated

to be the best pretreatment for the production of the

most favorable feed for PHA storage, by increasing

fermentation yield (from 25% to 36% at initial

concentration of 36.7 gCODL�1) and VFA concentra-

tion (from 9 to 13 gCODL�1) with respect to untreated

OMEs. Even though acids with an even number were

always predominant (acetic or butyric acids), OME

fermentation was effective in producing VFAs with an

odd number of carbon atoms (propionic and valeric

acids), which were absent in untreated OMEs. The latter

acids allow the formation of the HV monomer, which is

useful because an increasing HV content improves the

thermal and mechanical properties of the copolymer

P(HB-HV).

By using centrifuged and fermented OMEs, PHAs

were produced at an initial specific rate of

417mgCODgCOD�1 h�1 and the maximum HV con-

tent within the copolymer was 11% (on a molar basis).

HV production only occurred until propionic acid was

present in the medium. During the first hours of polymer

production, the storage yield was very high

(1mgPHAmgVFA�1 on a COD basis) indicating that

PHAs were also produced by carbon sources other than

VFAs. PHA production from centrifuged OME, with no

previous fermentation, showed to be lower (152 vs.

465 gCODL�1) and a less favorable composition (0%

HV vs. 11%) than that obtained from fermented OMEs.

Fermentation was therefore a very important step in

improving the overall performance of the process.

Even though less than maximum values were reported

in the literature for pure cultures, the final PHA

concentration was quite high (0.54 g PHAgVSS�1). This

aspect still deserves attention in order to maximize the

ARTICLE IN PRESSD. Dionisi et al. / Water Research 39 (2005) 2076–20842084

PHA content of biomasses and decrease extraction and

recovery costs.

Acknowledgments

This work was funded by the National Institute for

Safety and Health (Italy) under research contract B/13/

DIPIA/00. One of the authors (FC) is pleased to

acknowledge the financial support of the Spanish

Ministry of Education, Culture and Sports, making his

research stay at the University of Rome ‘‘La Sapienza’’

(Italy) possible.

References

Ahmad, A.L., Ismail, S., Bhatia, S., 2003. Water recycling from

palm oil mill effluent using membrane technology. Desali-

nation 157 (1–3), 87–95.

Andreozzi, R., Longo, G., Majone, M., Modesti, G., 1998.

Integrated treatment of olive oil mill effluents (OME): study

of ozonation coupled with anaerobic digestion. Water Res.

32 (8), 2357–2364.

Angelidaki, I., Ahring, B.K., 1997. Codigestion of olive oil mill

wastewaters with manure, household waste or sewage

sludge. Biodegradation 8 (4), 221–226.

APHA, AWWA, WPCF, 1992. Standard Methods for the

examination of water and wastewater, 18th ed. APHA,

Washington.

Beccari, M., Bonemazzi, F., Majone, M., Riccardi, C., 1996.

Interaction between acidogenesis and methanogenesis in the

anaerobic treatment of olive oil mill effluents. Water Res. 30

(1), 183–189.

Beccari, M., Carucci, G., Majone, M., Torrisi, L., 1999. Role of

lipids and phenolic compounds in the anaerobic treatment

of olive oil mill effluents. Environ. Technol. 20, 105–110.

Beccari, M., Petrangeli Papini, M., Majone, M., Torrisi, L.,

2001. Enhancement of anaerobic treatability of olive oil mill

effluents by addition of Ca(OH)2 and bentonite without

intermediate solid–liquid separation. Water Sci. Technol. 43

(11), 275–282.

Beccari, M., Carucci, G., Lanz, A.M., Majone, M., Petrangeli

Papini, M., 2002. Removal of molecular weight fractions of

COD and phenolic compounds in an integrated treatment of

olive oil mill effluents. Biodegradation 13, 401–410.

Borja, R., Banks, C.J., 1995. Comparison of an anaerobic filter

and an anaerobic fluidized bed reactors treating palm oil

mill effluent. Process Biochem. 30 (6), 511–521.

Carrasco, F., Dionisi, D., Majone, M., Petrangeli Papini, M.,

Carucci, G., Beccari, M., 2004. Los polihidroxialcanoatos:

plasticos biodegradables producidos por microorganismos.

Ing. Quım. (Madrid) 36 (414), 244–251.

Chin, K.K., Lee, S.W., Mohammad, H.H., 1996. A study of

palm oil mill effluent using a pond system. Water Sci.

Technol. 34 (11), 119–123.

Choi, J., Lee, S.Y., 2000. Economic considerations in the

production of poly(3-hydroxybutirate-co-hydroxyvalerate)

by bacterial fermentation. Appl. Microbiol. Biotechnol. 53,

646–649.

Cliffe, K.R., Patumsawad, S., 2001. Co-combustion of waste

from olive oil production with coal in a fluidised bed. Waste

Manage. 21 (1), 49–53.

Dionisi, D., Majone, M., Papa, V., Beccari, M., 2004.

Biodegradable polymers from organic acids by using

activated sludge enriched by aerobic periodic feeding.

Biotechnol. Bioeng. 85 (6), 569–579.

Galiatsatou, P., Metaxas, M., Arapoglou, D., Kasselouri-

Rigopoulou, V., 2002. Treatment of olive oil mill waste

water with activated carbons from agricultural by-products.

Waste Manage. 22 (7), 803–812.

Hassan, M.A., Shirai, Y., Kusubayashi, N., Abdul Karim,

M.I., Nakanishi, K., Hashimoto, K., 1996. Effect of organic

acids profiles during anaerobic treatment of palm oil mill

effluent on the production of PHA by Rhodobacter

sphaeroides. J. Ferment. Bioeng. 82 (2), 151–156.

Hassan, M.A., Shirai, Y., Kusubayashi, N., Abdul Karim,

M.I., Nakanishi, K., Hashimoto, K., 1997a. The production

of polyhydroxyalcanoate from anaerobically treated palm

oil mill effluent by Rhodobacter spheroides. J. Ferment.

Bioeng. 83 (5), 485–488.

Hassan, M.A., Shirai, Y., Umeki, H., Abdul Karim, M.I.,

Nakanishi, K., Hashimoto, K., 1997b. Acetic acid separa-

tion from anaerobically treated palm oil mill effluent for the

production of polyhydroxyalcanoate by Alcaligenes eutru-

phus. Biosci. Biotechnol. Biochem. 61, 1465–1468.

Lee, S.Y., 1996. Bacterial polyhydroxyalcanoates. Biotechnol.

Bioeng. 49 (1), 1–14.

Meesters, K.H.P., 1998. Production of poly(3-hydroxyalcano-

ates) from waste streams. Report of the Technical Uni-

versity of Delft, Delft (The Netherlands).

Reis, M.A.M., Serafim, L.S., Lemos, P.C., Ramos, A.M.,

Aguiar, F.R., Van Loosdrecht, M.C.M., 2003. Production

of polyhydroxyalkanoates by mixed microbial cultures.

Bioprocess Biosystems Eng. 25 (6), 377–385.

Saatci, Y., Arslan, E.I., Konar, V., 2003. Removal of total

lipids and fatty acids from sunflower oil factory effluent

UASB reactor. Biores. Technol. 87 (3), 269–272.

Salehizadeh, H., van Loosdrecht, M.C.M., 2004. Production of

polyhydroxyalcanoates by mixed culture: recent trends and

biotechnological importance. Biotechnol. Adv. 22, 261–279.

van Aalst-van Leeuwen, M.A., Pot, M.A., van Loosdrecht,

M.C.M., Heijnen, J.J., 1997. Kinetic modeling of poly(b-hydroxybutyrate) production and consumption by Para-

coccus pantotrophus under dynamic substrate supply.

Biotechnol. Bioeng. 55 (5), 773–782.

Yu, J., 2001. Production of PHA from starchy wastewater via

organic acids. J. Biotechnol. 86, 105–112.