Integration of Ozonation and an Anaerobic Sequencing BatchReactor (AnSBR) for the Treatment of Cherry Stillage

Pedro M. AÄ lvarez,* Fernando J. Beltran, and Eva M. Rodrıguez

Departamento de Ingenierıa Quımica y Energetica, Universidad de Extremadura, 06071 Badajoz, Spain

Cherry stillage is a high strength organic wastewater arising from the manufactureof alcoholic products by distillation of fermented cherries. It is made up of biorefractorypolyphenols in addition to readily biodegradable organic matter. An anaerobicsequencing batch reactor (AnSBR) was used to treat cherry stillage at influent CODranging from 5 to 50 g/L. Different cycle times were selected to test biomass organicloading rates (OLRB), from 0.3 to 1.2 g COD/g VSS‚d. COD and TOC efficiency removalshigher than 80% were achieved at influent COD up to 28.5 g/L but minimum OLRBtested. However, as a result of the temporary inhibition of acetogens and methanogens,volatile fatty acids (VFA) noticeably accumulated and methane production came to atransient standstill when operating at influent COD higher than 10 g/L. At theseconditions, the AnSBR showed signs of instability and could not operate efficiently atOLRB higher than 0.3 g COD/g VSS‚d. A feasible explanation for this inhibition is thepresence of toxic polyphenols in cherry stillage. Thus, an ozonation step prior to theAnSBR was observed to be useful, since more than 75% of polyphenols could beremoved by ozone. The integrated process was shown to be a suitable treatmenttechnology as the following advantages compared to the single AnSBR treatment wereobserved: greater polyphenols and color removals, higher COD and TOC removal ratesthus enabling the process to effectively operate at higher OLR, higher degree ofbiomethanation, and good stability with less risk of acidification.

IntroductionPublic concern for environmental problems arising

from food-processing industries is increasing, so produc-ers are making great efforts to introduce pollutionprevention and clean technologies into their processingoperations. Primary issue of concern is water pollutionbecause the food-processing industry usually generateslarge volumes of wastewater. Consequently, there iscurrently a great deal of interest in developing water andwastewater minimization technologies (i.e., closed loopand zero emission systems) and advanced wastewatertreatment practices (1).

The emerging sectors in the worldwide food industryinclude the production of alcohol, wines, and spirits fromfermentation of fruits and vegetables and further distil-lation of the fermented mash. Currently, the manufactureof alcoholic fruit drinks by distillation produces largeamounts of wastewater, which characteristics depend onthe raw material and the processes used in fermentationand distillation (2, 3). Particularly, our interest is focusedon cherry-based distilleries that produce from 8 to 15 Lof acidic wastewater (i.e., distillery slops or stillage) perliter of alcohol. Cherry stillage is characterized by highsolid content, high biodegradable dissolved organic mat-ter concentration, and a variable amount of inorganicmaterials (4). Besides the nonrecovered alcohol and cellyeast, the composition of the stillage includes a varietyof organic compounds such as products of fermentation,sugars residues, amino acids, pectins, tannins, and other

polyphenolic compounds from the feedstock (5, 6). Despitethe interest in mitigating the polluting effects of theseprocessing wastes through minimization options, the veryhigh organic content makes difficult direct reuse, recy-cling, or byproduct recovery. Moreover, final effluents ofthese practices are not generally suitable for dischargeand therefore must be treated to meet regulatory levels.The strategy for the choice of an appropriate treatmentsystem is not an easy task because it is based on multipledeciding-effects factors such as properties of the waste-water, quality requirements, flowrate of seasonal efflu-ent, pollution control regulations, climatic conditions, orlocation of the distillery. Nevertheless, as in most prac-tices dealing with other distillery wastewaters, physical/mechanical separation of the solid fraction of cherrystillage and the further application of biological methodsfor the treatment of the liquid fraction would likelyconstitute the most economically viable option (3).

In a series of previous research works (4, 7), dilutedcherry stillage (i.e., 1/80 dilution factor) was treated byan activated sludge system. The process was shown tobe effective in reducing levels of BOD and COD (namely,95% and 82%, respectively) though it could not ad-equately remove polyphenols (less than 35%). To bestachieve complete purification of diluted cherry stillage,the coupling of activated sludge and ozonation wasproved useful. The application of a preozonation stageimproved biodegradation rates, and a further ozonationprocess following aerobic biodegradation performed al-most complete elimination of biorefractory polyphenolsand overall COD. However, from a practical point of view,these schemes of treatment would be economically fea-

* To whom correspondence should be addressed. Tel/Fax: 34-924-289385. E-mail: pmalvare@unex.es.

1543Biotechnol. Prog. 2005, 21, 1543−1551

10.1021/bp049545d CCC: $30.25 © 2005 American Chemical Society and American Institute of Chemical EngineersPublished on Web 07/26/2005

sible only if the distillery is served by a nearby municipalwastewater treatment plant (MWTP) where dilution andneutralization of cherry stillage with domestic sewage isnot a drawback (8). In that case, the distillery must paya surcharge to have its wastewater further treated at theMWTP. To avoid this cost and/or to meet more stringentlimits beyond by-law requirements for the release ofwastewater to a MWTP, enabling reuse or direct streamdischarge from the distillery factory, an in-plant high-efficiency anaerobic system or a combination of anaerobicand aerobic processes are likely the best options (3).

Anaerobic digestion of stillages has potential advan-tages over aerobic treatment such as lower energy input,lower nutrients requirements, lower excess of sludge, andnet benefit of energy generation through the productionof a methane-containing biogas. Both mesophilic andthermophilic anaerobic digestion technologies have beensuccessfully applied to stillages from a variety of feed-stocks (3). The anaerobic method of treatment is, there-fore, expected to be more cost-effective for highly con-centrated cherry stillage compared to aerobic systems.The challenge, however, is to find a cost-effective anaero-bic reactor configuration and an easy way of operatingit. According to literature (9), most of the full-scaleanaerobic digestion systems treating various types ofdistillery wastewater are of the anaerobic upflow sludgeblanket (UASB) or expanded granular sludge bed (EGSB)types. However, given the seasonal nature and variabilityin quantity and quality of cherry stillage effluents, a moreflexible-operating reactor configuration is desirable. Inthis sense, the anaerobic sequencing batch reactor (AnS-BR), first developed by Dague and co-workers (10), is apromising technology. The AnSBR involves four consecu-tive steps accomplished in one vessel in a cyclic manner:fill, react, settle, and draw-off. The method of operationof such a system, which is described elsewhere (11),allows flexibility and quite simple operation without theneed of additional settling devices. Flexibility mainlyarises from its time-oriented nature that allows one toadjust operating conditions such as hydraulic retentiontime (HRT) or organic loading rate (OLR) by simplyresetting the feed volume or the sequence time for thestages of each cycle. AnSBRs have been successfullyapplied to the digestion of municipal sludge (12), landfillleachate (13), animal wastes (14), and a number of highstrength industrial wastewater containing large biologi-cal oxygen demand (BOD) and suspended solid (SS)concentration and/or inhibitory organic compounds(15-17).

For various types of distillery and other food-processingwastewaters, deficiencies in anaerobic digestion mayarise from the presence of certain compounds at inhibi-tory and/or toxic concentration levels. In such cases theanaerobic digestion needs complementing by aerobic and/or physicochemical processes to meet discharge limits.Particularly, the toxic effect of polyphenols and otherrefractory and inhibitory compounds present in distillerywastewater may be effectively avoided by applying achemical preoxidation using ozone that can transformpolyphenols and other complex structures into simpleroxygenated species of easier biodegradation (18, 19).

As part of our continuing work on cherry stillagetreatment, the objective of this research was the studyof its anaerobic digestion using an AnSBR and thecoupling of this biological method with a previous ozo-nation stage. The paper is primarily focused on overallprocess performance and stability of the biological stage.

Materials and Methods

Feed Wastewater. Cherry stillage was obtained froma distillery plant of the “Valle del Jerte” (Valdastillas,Caceres province, Spain). Mechanical separation of solidswas carried out by screening and centrifugation. Maincharacteristics of wastewater after SS separation havebeen previously published (4) and reproduced here inTable 1. This wastewater was stored frozen until its useto prevent biological activity. Before feeding the treat-ment system, the wastewater was thawed at ambienttemperature and preheated up to 20 °C.

Seed Sludge. Granular sludge from a full-scale UASBreactor treating paper mill wastewater was used asinoculum. The seed sludge had an average concentrationof total suspended solids (TSS) and volatile suspendedsolids (VSS) of 79.6 and 39.3 g/L, respectively.

Anaerobic Sequencing Batch Reactor. The anaer-obic treatment of cherry stillage was carried out in a lab-scale AnSBR. The digester was a jacketed glass-madebioreactor with a working volume of 1.8 L (ApplikonBiotechnology). The mixing of the reactor content wasprovided by an automated agitation system connected toa mechanical stirrer. Temperature, pH, and dissolvedoxygen level (DO) within the reactor were also continu-ously controlled within set point limits by a biocontrollersystem (Applikon Biotecnology, ADI 1030 model). Someperistaltic pumps (Masterflex, Cole Parmer) were usedfor various purposes as feeding influent wastewater fromthe storage tank, drawing-off treated effluent, and addingNaOH aqueous solution (15% v/v) in order to maintainthe pH at neutral conditions throughout the process.During settle and draw-off stages, nitrogen was injectedin the headspace to prevent pressure drop. The volumeof biogas produced from the digester was measured witha liquid displacement device containing an alkalinesolution. The system was run with the length of the cycleand the duration of each stage depending on the desiredreactor performance (11).

Start-Up of the AnSBR. The AnSBR was firstinoculated with 150 mL of seed sludge and 150 mL ofdistilled water to provide an initial VSS concentrationof about 3-4 g/L based on the total working volume ofthe reactor. The anaerobic consortia were fed with aneutralized cherry stillage/volatile fatty acids (VFA)mixture in a basal medium made up in distilled waterand containing yeast extract (0.2 g/L), KH2PO4 (170 mg/L), NH4Cl (37 mg/L), CaCl2‚2H2O (8 mg/L), MgSO4‚4H2O(9 mg/L), Na2S‚9H2O (10 mg/L), FeCl3‚6H2O (2 mg/L),CoCl2‚6H2O (2 mg/L), MnSO4‚H2O (0.5 mg/L), CuSO4‚5H2O (0.05 mg/L), ZnSO4‚7H2O (0.05 mg/L), H3BO3 (0.05mg/L), Na2MoO4‚2H2O (0.005 mg/L), Na2SeO4 (0.025 mg/L), NiSO4‚6H2O (0.05 mg/L), and EDTA (1 mg/L). Theamounts of cherry stillage and VFA in the feed provideda chemical oxygen demand (COD) of about 5 g/L. At theseconditions the COD/VSS ratio at the beginning of thereact phase was in the 1.25-1.75 g COD/g VSS range.

Table 1. Main Features of Cherry Stillage after SolidsSeparation

parameter unit value

pH 3.2-3.6COD g/L 145-180BOD g/L 100-140TOC g/L 70-90polyphenol g/La 2.0-2.5VFA g/Lb 6-18total acidity g/Lc 13-16

a As gallic acid. b As equivalent COD. c As CaCO3.

1544 Biotechnol. Prog., 2005, Vol. 21, No. 5

The percentage of cherry stillage in the feed was in-creased step by step throughout 2 months of the AnSBRstart-up period. The reactor content was continuouslymixed at 60 rpm during the fill and react stages, whileno agitation was provided during the settle and draw-off phases. Temperature and pH were controlled at 30 (2 and 7.2 ( 0.4 °C, respectively. Time conditions for thediscrete steps in each cycle were as follows: fill theeffective volume of the reactor by adding 1.5 L of feedduring 5 h; react during enough time to complete 80%COD removal at the least (react time varied between 1and 4 days); settle the sludge during 2 h; and draw-offthe volume added at the fill stage during 30 min. At theend of this last stage 0.3 L of sludge bed was left in thereactor.

Operation of the AnSBR. During the typical opera-tion of the AnSBR (after the start-up period) the feedcontained just cherry stillage (without external addedVFA), whether preozonated or not, diluted with distilledwater to have the required influent COD (ranged from 5to 50 g/L) and supplemented with NaHCO3 to providean alkalinity/COD ratio of 1:10, thus stabilizing pHaround 7.2. At the start of each cycle, the 0.3 L sludgebed contained about 25 g VSS/L. Agitation, temperature,and pH were controlled throughout the operation of theAnSBR at the same set point values as on the start-upperiod. Time conditions for the stages of the cycles werealso the same as those indicated above for the start-up,except for the react period that was in the range of 4-20days. Samples were withdrawn through a reactor sam-pling port during the react phase of each cycle to beanalyzed for COD, total organic carbon (TOC), and VFA.Additionally, TSS, VSS, and polyphenol concentrationswere measured in the effluent from the reactor. At theend of the draw-off stage, an idle phase was additionallyconsidered to (a) allow residual VFA within the bedsludge to be completely transformed into methane, (b)provide flexibility in starting a new cycle, and (c) add ordrain off sludge from the reactor in order to get aconvenient concentration of VSS in the digester for thenext cycle. Nevertheless, in any case the AnSBR was notleft idle for more than 24 h.

Methanogenic Activity Tests. Both the seed sludgefrom the full-scale UASB used as inoculum and that afterthe start-up of the AnSBR were characterized in termsof specific methanogenic activity (SMA) by performingbatch test assays. All assays were conducted at 35 °C.The basal medium was as described above containingessential macro- and micronutrients. The bioreactor wasloaded with a volume of sludge to provide an initialconcentration in the range of 4.5-6 g TSS/L (2-3.5 gVSS/L). Finally, a stock solution of VFA previouslyneutralized with NaOH (i.e., 0.4 g/L of acetate, 0.4 g/Lof propionate, and 0.4 g/L of butyrate) was added as feedto the methanogenic bacteria of the sludge. The volumeof biogas produced during the course of the assay wasfollowed by a liquid displacement technique. Chromato-graphic biogas analysis was also carried out to report themethane production. Liquid samples were withdrawnregularly during the course of the assays to be analyzedfor COD.

Ozonation. Ozonation was performed in semi-batchmode as pretreatment of undiluted cherry stillage priorto anaerobic treatment. The ozonation system basicallyconsisted of an ozone generator (Sander, model 307.1) anda 0.5-L stirred tank reactor. Experiments were conductedat room temperature and acidic pH (3.2-3.6). First, 0.4L of undiluted cherry stillage was charged into thereactor and the magnetic agitation system was turned

on (300 rpm) to provide good mixing conditions. Once theozone generation from pure dry oxygen was stabilizedwith the gas stream by-passing the reactor, a 25 L/h STPflow of gas was bubbled into the reactor through adiffuser located at the bottom. During the course of thereaction the concentrations of ozone in the continuousgas stream entering and leaving the reactor were moni-tored by an inline analyzer (Anseros Ozomat, modelGM19) to further calculate the used ozone dosages (7).Effluent samples were analyzed for COD, TOC, andpolyphenols. Ozonation experiments were consideredcompleted when a 75% reduction of polyphenols contentwas achieved at the least.

Analytical Methods. COD and TOC were analyzedon filtered samples (0.45 µm pore diameter) following thedichromate method with a cuvette test (Dr. Lange,LCK114) and with a Dohrmann DC190 analyzer, respec-tively. The respirometric method was used to measureBOD (OxiTop WTW). Polyphenol concentration wasdetermined by the Folin Ciocalteau method using gallicacid as the standard as described elsewhere (20). Acidity,alkalinity, TSS, and VSS analyses were according toprocedures outlined in Standard Methods (21). Methaneconcentration in the biogas was analyzed by a gaschromatography system provided with a thermal con-ductivity detector (Konik, model 3000 HR-GC). VFA weremeasured on filtered (0.45 µm pore diameter) acidifiedsamples (up to pH 2 with o-phosphoric acid) with aHewlett-Packard 5890 GC equipped with a flame ioniza-tion detector. A capillary Supelco column (Vocol phasestationary; 105 m long and 0.53 mm i.d.) was used. High-purity helium was the carrier gas at 100 cm3/min.Injector and detector block temperatures were kept at250 °C. The oven temperature program was as follows:110 °C for 1 min, heat at a rate of 3 °C/min to 110 °Cand then hold for 1 min, heat at a rate of 3 °C/min to145 °C and then hold for 2 min.

Results and Discussion

Specific Metanogenic Activity (SMA) of theSludge. The SMA of the seed sludge used for inoculationof the AnSBR (i.e., nonacclimated sludge) and that afterthe AnSBR start-up (i.e., sludge acclimated to cherrystillage) were first studied. Figure 1 shows the averagecumulative volume of biogas during the course of twoassays taken as examples. As seen, methane productionwas almost independent of whether the sludge used was

Figure 1. Biogas production during SMA tests: (9) nonaccli-mated sludge; (b) acclimated sludge.

Biotechnol. Prog., 2005, Vol. 21, No. 5 1545

acclimated to cherry stillage or not. Moreover, the biogashad a similar composition (72-78% methane) in all ofthe assays. The quick initial biogas production and theabsence of steps in the methane production curves areindicative of the absence of an initial lag phase (i.e.,acclimation) and inhibition effects, respectively.

Figure 2 shows the evolution of the COD measuredduring the course of the batch bioassays. Theoretically,assuming that the VFA mixture is stoichiometricallyconverted into methane, each milliliter of methaneproduced is equivalent to 2.86 mg of COD. Accordingly,Figure 2 also presents the calculated COD profile con-sidering just only the substrate removed via methaneproduction (i.e., CODCH4). As seen, at any time over theexperimental period, CODCH4 was relatively close toactual COD, which validates the choice of the substrateused for the purpose of this test.

The well-known kinetic model of Monod was assumedto calculate the SMA according to

where the maximum specific activity (SMA) is definedas

The differential eq 1 can be simplified in a number ofcases (22). Particularly, when a zero order kinetics isobserved (i.e., S , KS and X0 . YX/S(S0 - S)), eq 1 can beeasily simplified and solved to yield

In our study, conditions of the tests were selected inorder to fulfill eq 3 at the beginning of the assays (23).Accordingly, taking the substrate concentration (S) asCODCH4 and the microorganism concentration (X0) as theVSS at the start of the assay, from the slopes of thestraight lines of Figure 2 the SMAs of the acclimated andnonacclimated sludge were calculated to be 0.20 and 0.22g CODCH4/g VSS‚d, respectively. Accordingly, the seedused as inoculum had a SMA somewhat low but withinthe range of most industrial and laboratory anaerobicdigesters (0.1-1 g CODCH4/g VSS‚d). Moreover, the SMAvalue was not modified appreciably after the start-up ofthe AnSBR treating cherry stillage.

Performance and Stability of the AnSBR Treat-ing Nonozonated Cherry Stillage. After the start-up,the functioning of the AnSBR treating nonozonatedcherry stillage was monitored over a 5 month period. Thesystem was operated at four increasing influent CODfrom 5.5 to 28.5 g/L (runs 1-7). Cycle time was variedfrom one run to another aimed at obtaining preselectedOLRB’s in the range of 0.3-0.6 g COD/g VSS‚d. Thisrange was chosen to exploit the methanogenic activityof the biomass to the maximum in accordance with theSMA results shown above. Table 2 summarizes operatingconditions of these runs and overall performance in termsof COD, TOC, and polyphenol (PPh) removal efficienciesand methane yield (that is, the percentage of CODremoved transformed into methane). Although every runcomprised at least two cycles, the reported results arethose of the last one in order to minimize the transienteffects of the step loading increases. The results in Table2 show that effective purification of cherry stillage wasattained when the AnSBR was operated at the lowestOLRB (runs 1, 2, 4, and 6). Thus, regardless of theinfluent COD, at an OLRB of 0.3 g COD/g VSS‚d, CODremoval efficiency and mineralization of the organicmatter (i.e., TOC removal) were achieved to a highextent. At this OLRB methane yield was greater than 60%(i.e., 198.3 mL CH4 per g of added COD as average figure)but below the average value reported for the mesophilicanaerobic digestion of other types of stillage, typicallyabout 250 mL CH4/g COD (3). At increasing OLRB from

Table 2. Conditions Examined and Performance during Operation of the AnSBR

runinfluent

wastewaterainfluent

COD (g/L)cycle

time (d)HRT(d)

OLRB(g COD/g VSS‚d)

CODremoval

(%)

TOCremoval

(%)

methaneyield(%)

PPhremoval

(%)

effluentVFA

(g COD/L)

effluentpropionate(g COD/L)

1 CS 5.5 4.5 5.4 0.3 90.0 90.2 71.1 16.7 0.13 0.122 CS 10.0 7.0 8.4 0.3 84.4 82.4 61.4 17.4 1.00 0.833 CS 10.0 5.0 6.0 0.4 72.4 68.0 60.9 17.3 2.85 2.074 CS 16.5 13.0 15.6 0.3 87.1 87.6 63.5 35.3 0.93 0.765 CS 16.5 7.0 8.4 0.5 65.9 60.3 58.6 18.4 3.86 3.426 CS 28.5 19.0 22.8 0.3 92.7 91.6 67.4 43.2 0.82 0.547 CS 28.5 8.5 10.2 0.6 69.1 60.6 60.4 20.3 7.98 7.328 OCS 16.5 13.0 15.6 0.3 95.6 89.4 76.2 76.7b 1.14 0.599 OCS 16.5 7.0 8.4 0.5 87.0 84.7 72.0 74.3b 1.86 1.1210 OCS 21.5 8.5 10.2 0.5 89.3 87.7 74.4 80.0b 1.38 0.9811 OCS 48.5 8.5 10.2 1.2 52.4 47.6 69.9 83.2b 17.6 13.3

a CS: cherry stillage. OCS: ozonated cherry stillage. b Removal achieved within the integrated process: ozonation and AnSBR treat-ment.

Figure 2. Normalized residual COD during SMA tests: (9)COD, nonacclimated sludge, (b) COD, acclimated sludge; (0)CODCH4, nonacclimated sludge, (O) CODCH4, acclimated sludge.Lines are linear fits according to eq 3: (s) nonacclimated sludge;(---) acclimated sludge.

rs ) - dSdt

)µmax

YX/S

S‚[X0 + YX/S(S0 - S)]

(KS + S)(1)

SMA )µmax

YX/S(2)

SS0

) 1 -SMA‚X0

S0t (3)

1546 Biotechnol. Prog., 2005, Vol. 21, No. 5

0.3 g COD/g VSS‚d (runs 3, 5, and 7) the COD and TOCremoval efficiencies decreased. It is interesting to noticethat in these experiments, performed at OLRB from 0.4to 0.6 g COD/g VSS‚d, the conversion of TOC wassomewhat lower than that of COD, likely because of theaccumulation of intermediates (i.e., VFA). A fact insupport of this argument is that the methane yieldsobserved were lower than those achieved in runs per-formed at the same influent COD but lower OLRB (i.e,runs 3, 5, and 7 versus 2, 4, and 6, respectively). Thus,methane production was below 150 mL CH4/g COD inruns 3, 5, and 7. It is worth noting in Table 2 that theefficiency of polyphenol removal was rather low in allruns from 1 to 7 regardless of the OLRB applied and thatit was HRT-dependent. This is likely due to the slowkinetics of anaerobic digestion of polyphenols, which asa rule show low initial degradation rates (24). Hence, asthe HRT was risen the anaerobic microbial consortia ofthe digester had time for the assimilation of morepolyphenols.

To further investigate the reasons for the low efficiencyof the AnSBR when operating at OLRB higher than 0.3g COD/g VSS‚d, the evolutions of accumulated methaneand residual COD during the course of a cycle of each offour runs carried out at different influent COD (runs 1,2, 4, and 6) were recorded as shown in Figure 3. It canbe seen that the anaerobic bioculture had the ability toremove COD at the beginning of every cycle, showing noevidence of initial lag induction phase. However, as theinfluent strength was increased, the COD of sampleswithdrawn from the reactor became steadily earlier.Particularly, from the stepped shape of the COD-timeand methane-time profiles of runs 4 and 6 it is clearlyseen that the first period of rapid COD removal andmethane production was followed by a phase of inactivity,after which the COD removal and methane productionrate were recovered. Vertical dotted lines denote the timeconditions corresponding to an applied OLRB of 0.3 gCOD/g VSS‚d. As seen, at the left side of these linesresidual COD is rather large, specially for runs 4 and 6.

This behavior hints that the temporary inhibition ofmethanogenic bacteria rendered unfit the AnSBR toadequately treat cherry stillage at OLRBs above 0.3 gCOD/g VSS‚d.

Inhibition of methanogenesis is usually characterizedby the accumulation of VFA, which in turn may causethe failure of the digester because of excessive acidifica-tion. Accordingly, VFA concentration has been usuallyused as indicator of the reactor stability (25). Figure 4shows the evolution of some specific VFA (i.e., acetate,propionate, butyrate, isobutyrate, valeriate, and isovale-riate) as well as overall VFA concentration measured asequivalent COD (i.e., CODVFA) in runs 1, 3, 5, and 7. Allfour graphs show that acetate and propionate were themajor constituents of VFA, whereas concentrations ofisoacids were the lowest. From a qualitative point of view,the evolution of CODVFA during the course of all runsshowed two stages: from the start of the cycle CODVFAincreased sharply up to reach a maximum and thendeclined more or less parallel to the COD curve. It is alsointeresting to note that in Figure 4 the maximum CODVFAcoincided more or less with a sharp decrease in theoverall COD removal rate, that is, when the inhibitionprocess started to develop.

In runs 5 and 7, where inhibition of methane produc-tion was more evident from COD profiles of Figure 4, theconcentration of acetate at the time corresponding to themaximum CODVFA was much beyond the limit of 0.8 g/Lproposed for a good equilibrium between the syntrophicgroups of the anaerobic consortia (25). In fact, at this time(1.3 days for run 5 and 2 days for run 7) the systemexhibited signs of instability and addition of NaOH wasnecessary to avoid acidification (i.e., pH drop below 6.8)due to the massive accumulation of VFA. In runs 1 and3, however, the initial alkalinity of the system was largeenough to maintain neutral pH without adding extraalkali. In all runs, from the time corresponding to themaximum CODVFA up to the end of the cycle, the pH waskept at 7.2-7.5 as a result of the auto-buffering capacityof the system. Therefore, it is presumed that hydrolysisand acidogenesis processes were already almost com-pleted at the time of maximum CODVFA. In fact, thedegree of acidification, which can be defined as thepercentage of influent COD converted into methane andVFA, was 81.1%, 78.3%, 78.7%, and 83.4% for runs 1, 3,5, and 7, respectively. The remaining up to 100% is likelydue to the refractory COD and that used for cell growth.

The second stage of CODVFA curves in Figure 4 ischaracterized in all runs by a decrease that coincides withthe depletion of the overall COD. This indicates that theAnSBR performance was limited by the rate of VFAconversion. It can be observed that while in run 1 a rapidand complete removal of VFA was achieved, in runs 3,5, and 7 the effluents were largely composed of the VFAproduced in the reactor (mainly propionate) that werenot degraded within the cycle time. The build-up ofpropionate concentration was likely a result of theinhibition of syntrophic propionate degraders, as hasbeen observed in a number of anaerobic systems loadedwith inhibitors as, for example, phenolic compounds, longchain fatty acids, or linear alkylbenzene sulfonates (26-28). Nevertheless, the inhibition was not permanent sincewhen the cycle was long enough, almost complete re-moval of VFA was attained. In fact, as shown in Table 2,the concentration of overall VFA and propionate in theeffluents of runs 2, 4, and 6 (see Table 2 to note that HRTof runs 2, 4, and 6 was greater than that correspondingto runs 3, 5, and 7, respectively) were in all cases lowerthan 1 and 0.85, respectively. Moreover, the concentra-

Figure 3. Evolution of accumulated methane volume (solidsymbols) and COD (open symbols) during the operation of theAnSBR treating nonozonated cherry stillage at different influentCOD: (9, 0) run 1, CODi ) 5.5 g/L; (b, O) run 2, CODi )10.0 g/L; (2, 4) run 4, CODi ) 16.5 g/L; (1, 3) run 6, CODi )28.5 g/L. See Table 2 for other specific conditions of each run.

Biotechnol. Prog., 2005, Vol. 21, No. 5 1547

tion of acetate in the effluents of all of these runs wasnegligible, which is in agreement with the findings ofPullammanappallil et al. (29). Acordingly, it can beconcluded that propionate did not inhibit aceticlasticmethanogenesis to a high extent but might be a cause ofthe slow degradation of propionate into acetate.

Effect of Preozonation of Cherry Stillage on thePerformance and Stability of the AnSBR. Experi-mental facts described in the previous section point outthe high organic load as responsible for a type ofreversible inhibition of the anaerobic biodegradation ofcherry stillage. As reported for other food-processingwastewaters (30, 31), this inhibition could arise from thedetrimental effect of tannins and other polyphenols(noticeably present in cherry stillage as can be seen inTable 1). These substances can inhibit or slow thekinetics of the acetogenic and methanogenic bacteria,although other factors might also have major impacts.

To ascertain whether polyphenols of cherry stillageaffected negatively the process performance and stabilityof the AnSBR, two series of runs were carried out onpreozonated cherry stillage. As it is well-known, ozonecan react with biorefractory polyphenols, transformingthem into oxygenated low molecular weight products thatare more readily biodegradable (32). Accordingly, thetarget of preozonation in this work was the specificremoval of polyphenols of cherry stillage rather than theoverall COD reduction. To best achieve this objective thepreozonation treatment was carried out on undilutedcherry stillage at pH 3.2-3.6, thus favoring selectivedirect reactions between ozone and phenolic compounds(33). This proved to be satisfactory since more than 75%

polyphenols removal was achieved with a used ozone doseas low as 25 mg of ozone per gram of initial COD.Operating at this dose, the COD removal efficiency waswell below 5%. Another beneficial effect of ozonation wasthe partial discoloration of the dark brown color of thecherry stillage, which likely arises from the polymeriza-tion of phenolic compounds and tannins during storage(34).

Conditions of four AnSBR experiments on preozonatedeffluents (runs 8-11) are indicated in Table 2 togetherwith their process performance. First, runs 8 and 9 werecarried out at the same conditions as runs on non-preozonated cherry stillage 4 and 5, respectively. Tobetter compare the AnSBR performance with cherrystillage treated and not treated with ozone, Figure 5shows the results graphically. The improvement inprocess performace induced by ozonation is clear fromthe greater COD, TOC, and PPh removals attained inruns 8 and 9 compared to runs 4 and 5, respectively. Thedegree of methanation was also significantly improvedby means of ozonation, as shown by the large methaneyield achieved in runs 8 and 9 (i.e., 233.1 mL per g ofinfluent COD as average figure), which gives an addedvalue to the integrated process from the energy recoverypoint of view. Moreover, it is interesting to note fromFigure 5 how the goal of preozonation was more out-standing when the AnSBR was operated at an OLRB of0.5 g COD/g VSS‚d. This leads one to think that pre-ozonation might suppress to an extent the inhibition ofthe biological step.

To further investigate this last point a second seriesof experiments was carried out to test the effectiveness

Figure 4. COD and VFA profiles during cycles of the AnSBR treating non-preozonated cherry stillage at different OLRB (g COD/gVSS‚d). Run 1: OLRB ) 0.3. Run 3: OLRB ) 0.4. Run 5: OLRB ) 0.5. Run 7: OLRB ) 0.6. (9) COD; (0) CODVFA; (b) acetate; (2)propionate; (1) butyrate; ([) isobutyrate; (solid right triangle) valerate; (solid left triangle) isovalerate. See Table 2 for other specificconditions of each run.

1548 Biotechnol. Prog., 2005, Vol. 21, No. 5

of the AnSBR in treating ozonated cherry stillage in theOLRB range of 0.5-1.2 g COD/g VSS‚d while keeping theHRT at 10.2 days (runs 10 and 11). As seen in Table 2,COD and TOC removal efficiencies achieved in run 10were quite satisfactory bearing in mind the high strengthinfluent wastewater (i.e., 21.5 g/L COD). The percentageof COD and TOC removals in run 11 were, however, verylimited even though the methane yield approached 70%.The incomplete purification of cherry stillage in run 11

was likely due to the low HRT rather than to methano-genic inhibition, as inferred from Figure 6 where theevolutions of accumulated methane and residual CODthrough a cycle of runs 10 and 11 are plotted against thecycle time. It can be observed that neither the disap-pearance of COD nor the methane production werestopped at any moment of the cycles, reflecting thesuppression of the temporary inhibition by means ofozonation.

Figure 7 shows the evolution of VFA during a cycle ofruns 10 and 11. As in experiments with non-preozonatedcherry stillage, the main metabolites of acidification wereacetate and propionate (the first exceeding the limit of0.8 g/L) and the profiles of CODVFA resemble those ofFigure 4. However, a more detailed analysis of thesefigures discloses some indicators of the AnSBR processstability as summarized in Table 3. As pointed out before,the critical point for inhibition is that where CODVFAreachs the highest value. Indicators of process stabilityat this point are the concentration of overall VFA andparticularly of acetate and propionate. As seen in Table3, the overall concentration of VFA, expressed as CODVFAand normalized with respect to influent COD, was as a

Table 3. Some Indicators of the AnSBR Process Stability Treating Cherry Stillage

runa CODVFA/CODib CODacetate/CODi

b CODpropionate/CODib C3/C2

b (C3/C2)effluent

5 0.50 0.23 0.20 0.63 59.47 0.53 0.21 0.27 0.89 32.7

10 0.42 0.18 0.19 0.77 11.411 0.41 0.19 0.23 0.59 4.1

a For run conditions see Table 2. b Values corresponding to the reaction times that lead to maximal VFA concentration.

Figure 5. Comparison of AnSBR process performace treatingnon ozonated (runs 4 and 5) and ozonated cherry stillage(runs 8 and 9) at various OLRB (g COD/g VSS‚d). Runs 4 and 8:OLRB ) 0.3. Runs 5 and 9: OLRB ) 0.5. See Table 2 for otherspecific conditions of each run.

Figure 6. Evolution of accumulated methane volume (solidsymbols) and COD (open symbols) during the operation of theAnSBR treating preozonated cherry stillage at different influentCOD: (9, 0) run 10, CODi ) 21.5 g/L; (b, O) run 11, CODi )48.5 g/L. See Table 2 for other specific conditions of each run.

Figure 7. COD and VFA profiles during cycles of the AnSBRtreating preozonated cherry stillage at different OLRB (g COD/gVSS‚d). Run 10: OLRB ) 0.5. Run 11: OLRB ) 1.2. (9) COD;(0) CODVFA; (b) acetate; (2) propionate; (1) butyrate; ([)isobutyrate; (solid right triangle) valerate; (solid left triangle)isovalerate. See Table 2 for other specific conditions of each run.

Biotechnol. Prog., 2005, Vol. 21, No. 5 1549

rule lower in runs with preozonated wastewater. Thiswas likely due to the enhanced initial methanogenicactivity that prevented a massive accumulation of VFAat the end of the acidification phase. In fact, initial biogasformation rates were calculated to be 638.2, 827.2, 740.5,and 1047.5 mL CH4/g VSS‚d in runs 5, 7, 10, and 11,respectively. In connection with the lowering of CODVFA/CODi ratio induced by ozonation, it is interesting toremark that, contrary to runs 5 and 7, in runs withpreozonated cherry stillage the pH was always above 6.8without external addition of alkali. Thus, the initialalkalinity plus the carbon dioxide produced as a resultof organic matter mineralization that remained in solu-tion as bicarbonate was enough to avoid the risk of failureby acidification. Also in Table 3 it can be seen that theconcentrations of acetate and propionate (again normal-ized to the influent COD) were somewhat lower inexperiments with preozonated cherry stillage. Moreover,the relationship between propionate (C3) and acetate (C2)concentrations at the critical point was maintained at agood level in all the experiments regardless of cherrystillage preozonation, bearing in mind the limit of 1.4reported by Hill et al. (25) for this ratio in order toprevent digester failure. However, the C3/C2 ratio in-creased drastically from the critical point up to the endof the cycle in experiments performed on non-preozonatedcherry stillage. Thus, in the effluents of runs 5 and 7 thisratio went much beyond the above-mentioned limit.Contrary, in experiments with preozonated cherry still-age, a much lower accumulation of propionate wasobserved, which in turn enhanced the overall kinetics ofthe process as it reduced the effect of the substrateinhibition process (35). Nevertheless, accumulation ofpropionate in runs 11 and 12 was also greater thandesirable. This suggests that factors other than inhibitorypolyphenols might also have major impacts on processperformance and reactor stability. Among these factors,the very low fill to cycle time ratio used for the operationof the AnSBR (36) and/or simply the reactor stressprovoked by high load of readily biodegradable organicmatter should be highlighted.

Conclusions

The start-up of an AnSBR for the treatment of highstrength cherry stillage could be easily completed in 2months using a granular sludge from a full-scale UASBreactor treating paper mill wastewater as seed andincreasing the OLRB gradually. After start-up, the systemwas capable of effectively treating cherry stillage atinfluent COD lower than 10 g/L and OLRB of 0.3 g COD/gVSS‚d, operating in a stable way while meeting treat-ment objectives. At higher influent COD, however, theAnSBR showed signs of instability and COD and TOCremoval efficiencies significantly decreased at increasingOLRB’s. This poor performance is likely to be attributedmainly to the temporary inhibition of acetogenic andmethanogenic syntrophic anaerobic groups that mayarise mainly from the presence of toxic tannins and otherpolyphenols in cherry stillage.

A low dosage of ozone as pretreatment of cherry stillageprior to AnSBR operation has been demonstrated asuseful to ameliorate both the overall process performanceand the stability of the AnSBR. Polyphenols concentra-tion and color were reduced to a high extent by means ofozonation, though COD was practically unchanged withthe pretreatment. After ozonation, the AnSBR was ableto operate in a stable way at an OLRB range of 0.3 to1.2 g COD/g VSS‚d and the methane yield increased

significantly in comparison with that achieved within thetreatment of nonozonated cherry stillage. Although theinhibitory effects were greatly reduced by means ofozonation, propionate accumulated in the reacting me-dium, which negatively affected the process kineticsbecause of the inhibition of the propionate degradationinto acetate.

AcknowledgmentThe authors wish to express their thanks to the CICYT

of Spain and the European Commission for the financialsupport of this investigation under grant 1FD97-0087.

References and Notes(1) Sekoulov, I. Sustainable development of wastewater treat-

ment strategies for the food industries. Water Sci. Technol.2002, 45, 315-320.

(2) Sheehan, G. J.; Greenfield, P. Utilisation, treatment anddisposal of distillery wastewater. Water Res. 1980, 14, 257-277.

(3) Wilkie, A. C.; Riedesel, K. J.; Owens, J. M. Stillage char-acterization and anaerobic treatment of ethanol stillage fromconventional and cellulosic feedstocks. Biomass Bioenerg.2000, 19, 63-102.

(4) Beltran, F. J.; AÄ lvarez, P. M.; Rodrıguez, E. M.; Garcıa-Araya, J. F.; Rivas, J. Treatment of high strength distillerywastewater (cherry stillage) by integrated aerobic biologicaloxidation and ozonation. Biotechnol. Prog. 2001, 17, 462-467.

(5) Tsao, R.; Yang, R. Optimization of a new mobile phase toknow the complex and real polyphenolic composition: towardsa total phenolic index using high-performance liquid chro-matography. J. Chromatogr. A 2003, 1018, 29-40.

(6) Mozetic, B.; Trebse, P. Identification of sweet cherry antho-cyanins and hydroxycinnamic acids using HPLC coupled withDAD and MS detector. Acta Chim. Slov. 2004, 51, 151-158.

(7) Beltran, F. J.; AÄ lvarez, P. M.; Rodrıguez, E. M.; Rivas, J.Incidence of an ozonation stage on the treatment of cherrystillage by activated sludge. Ozone: Sci. Eng. 2004, 26, 257-266.

(8) Gruttner, H. Regulating industrial wastewater dischargedto public wastewater treatment plants-A conceptual ap-proach. Water Sci. Technol. 1997, 36, 25-33.

(9) Rajeshwari, K. V.; Balakrishnan, M.; Kansal, A.; Lata, K.;Kishore, V. V. N. State-of-the-art of anaerobic digestiontechnology for industrial wastewater treatment. RenewableSustainable Energ Rev. 2000, 4, 135-156.

(10) Dague, R. R.; Habben, C. E.; Pidaparti, S. R.; Initial studieson the anaerobic sequencing batch reactor. Water Sci. Tech-nol. 1992, 26, 2429-2432.

(11) Ruiz, C.; Torrijos, M.; Sousbie, P.; Martinez, J. L.; Moletta,R. The anaerobic SBR process: basic principles for design andautomation. Water Sci. Technol. 2001, 43, 201-208.

(12) Chang, D.; Hur, J. M.; Chung, T. H. Digestion of municipalsludge by anaerobic sequencing batch reactor. Water Sci.Technol. 1994, 30, 161-170.

(13) Timur, H.; Ozturk, I. Anaerobic sequencing batch reactortreatment of landfill leachate. Water Res. 1999, 33, 3225-3230.

(14) Angenent, L. T.; Sung, S. W.; Raskin, L. Methanogenicpopulation dynamics during startup of a full-scale anaerobicsequencing batch reactor treating swine waste. Water Res.2002, 36, 4648-4654.

(15) Masse, D. I.; Masse, L.; Verville, A.; Bilodeau, S. The start-up of anaerobic sequencing batch reactors at 20 °C and 25°C for the treatment of slaughterhouse wastewater. J. Chem.Technol. Biotecnol. 2001, 76, 393-400.

(16) Ruiz, C.; Torrijos, M.; Sousbie, P.; Martinez, J. L.; Moletta,R.; Delgenes, J. P. Treatment of winery wastewater by ananaerobic sequencing batch reactor. Water Sci. Technol. 2002,45, 219-224.

(17) Ribarova, I.; Topalova, J.; Ivanov, I.; Kozuharov, D.;Dimkov, R.; Cheng, C. Anaerobic sequencing batch reactor

1550 Biotechnol. Prog., 2005, Vol. 21, No. 5

as initiating stage in complete pentachlorophenol biodegrada-tion. Water Sci. Technol. 2002, 46, 565-569.

(18) Martın, M. A.; Raposo, F.; Borja, R.; Martın, A. Kineticstudy of the anaerobic digestion of vinasse pretreated withozone, ozone plus ultraviolet light, and ozone plus ultravioletlight in the presence of titanium dioxide. Process Biochem.2002, 37, 699-706.

(19) Santos, M. M. A.; Bocanegra, J. F. L.; Martın, A. M.; Garcıa,I. G. Ozonation of vinasse in acid and alkaline media. J.Chem. Technol. Biotechnol. 2003, 78, 1121-1127.

(20) Garcıa-Garcıa, I.; Bonilla-Venceslada, J. L.; Jimenez-Pena,P. R.; Ramos-Gomez, E. Biodegradation of phenol compoundsin vinasse using aspergillus terreus and geotrichum candi-dum. Water Res. 1997, 31, 2005-2011.

(21) APHA. Standard Methods for the Examination of Waterand Wastewater; Eaton, A. D., Clesceri, L. S., Greenberg, A.E., Eds.; American Public Health Association: Washington,DC, 1995.

(22) Simkins, S.; Alexander, M. Models for mineralizationkinetics with the variables of substrate concentration andpopular density. Appl. Environ. Microbiol. 1984, 47, 1299-1306.

(23) Soto, M.; Mendez, R.; Lema J. M. Methanogenic andnonmetanogenic activity tests. Theroretical basis and experi-mental set up. Water Res. 1993, 27, 1361-1376.

(24) Borja, R.; Alba, J.; Banks, C. J. Impact of the main phenoliccompounds of olive mill wastewater (OMW) on the kineticsof acetoclastic methanogenesis. Process Biochem. 1997, 32,121-133.

(25) Hill, D. T.; Cobb, S. A.; Bolte, J. P. Using volatile fatty acidrelationships to predict anaerobic digester failure. Trans.ASAE 1987, 30, 496-501.

(26) Fang, H. H. P.; Chan, O. C. Toxicity of phenol towardsanaerobic biogranules. Water Res. 1997, 31, 2229-2242.

(27) Kim, S. H.; Han, S. K.; Shin, H. S.Kinetics of LCFAinhibition on acetoclastic methanogenesis, propionate deg-radation and beta-oxidation. J. Environ. Sci. Health, Part A2004, 39, 1025-1037.

(28) Gavala, H. N.; Ahring, B. K. Inhibition of the anaerobicdigestion process by linear alkylbenzene sulfonates. Biodeg-radation 2002, 13, 201-209.

(29) Pullammanappallil, P. C.; Chynoweth, D. P.; Lyberatos,G.; Svoronos, S. A. Stable performance of anaerobic digestionin the presence of a high concentration of propionic acid.Bioresour. Technol. 2001, 78, 165-169.

(30) Borja, R.; Martın, A.; Maestro, R.; Luque, M.; Duran, M.M. Enhancement of the anaerobic digestion of wine distillerywastewater by the removal of phenolic inhibitors. Bioresour.Technol. 1993, 45, 99-104.

(31) Raposo, F.; Borja, R.; Sanchez, E.; Martın, M. A.; Martın,A. Inhibition kinetics of overall substrate and phenolicsremovals during the anaerobic digestion of two-phase olivemill effluents (TPOME) in suspended and immobilized cellreactors. Process Biochem. 2003, 39, 425-435.

(32) Amat, A. M.; Arques, A.; Beneyto, H.; Garcıa, A.; Miranda,M. A.; Segui, S. Ozonisation coupled with biological degrada-tion for treatment of phenolic pollutants: a mechanisticallybased study. Chemosphere 2003, 53, 79-86.

(33) Beltran, F. J.; Garcıa-Araya, J. F.; Rivas, J.; Alvarez, P.M.; Rodrıguez, E. Ozone remediation of some phenol com-pounds present in food processing wastewater. J. Environ.Sci. Health, Part A 2000, 35, 681-699.

(34) Assas, N.; Ayed, L.; Marouani, L.; Hamdi, M. Decolorizationof fresh and stored-black olive mill wastewaters by Geotri-chum candidum. Proccess Biochem. 2002, 38, 361-365.

(35) Mosche, M.; Jordening, H. J. Comparison of differentmodels of substrate and product inhibition in anaerobicdigestion. Water Res. 1999, 33, 2545-2554.

(36) Shizas L.; Bagley, D. M. Improving anaerobic sequencingbatch reactor performance by modifying operational param-eters. Water Res. 2002, 36, 363-367.

Accepted for publication June 27, 2005.

BP049545D

Biotechnol. Prog., 2005, Vol. 21, No. 5 1551