j our na l ho me page: www.elsev ier .com/ locate /ca t tod
olar photo-Fenton treatment of winery effluents in a pilothotocatalytic reactor
. Velegrakia,∗, D. Mantzavinosb
Department of Environmental Engineering, Technical University of Crete, Polytechneioupolis, GR-73100 Chania, GreeceDepartment of Chemical Engineering, University of Patras, Caratheodory 1, University Campus, GR-26504 Patras, Greece
r t i c l e i n f o
rticle history:eceived 17 January 2014eceived in revised form 30 May 2014ccepted 2 June 2014vailable online xxx
eywords:
a b s t r a c t
A pilot-scale solar Fenton process has been applied for the treatment of winery wastewater collected dur-ing the vinification period. The importance of the experimental variables was investigated at lab-scaleexperiments through the application of experimental design methodology. The pilot-scale study was con-ducted on a pilot CPC photocatalytic reactor under natural solar irradiation. The results show that at lowcatalyst dose (i.e. [Fe2+] = 5 mg L−1) mineralization (i.e. ca. 50%) is dependent on the oxidant consumption(i.e. 500 mg L−1), irrespective of the excess oxidant present; however, shorter reaction times are required
hoto-Fentoninery wastewater
xperimental designolar photocatalysis
under excess H2O2, indicating higher reaction rates due to higher availability of oxidant molecules in thebulk liquid. Increasing the catalyst dose enhances the reaction rate due to higher H2O2 decompositionand HO• production. This is corroborated with the lower H2O2 consumption (i.e. 1270 mg L−1) occurringat low catalyst, signifying, however, a more effective use of the oxidant (i.e. less oxidant is required toachieve similar mineralization).
Wine industry is an ever growing sector of the food indus-ry worldwide. In 2011 the world wine production exceeded6,600,000 t noting a 2.9% increase compared to the respective pro-uction of 2008 [1]. Wine industry has traditionally been subjecto a lesser amount of regulatory attention when compared to otherndustries e.g. chemicals and mining, with obvious environmen-al impacts; however, there are several environmental issues withhich wine producers have to contend [2,3], as the quality of theirroduct is directly linked to the qualitative characteristics of theaw materials (i.e. grapes), which in turn reflect to the soil and wateruality of viticulture practices.
Wine making is accompanied by various processes that com-ence immediately after grape harvesting; destemming, crushing
nd primary (alcoholic) fermentation are followed by cold stabi-ization and secondary (malolactic) fermentation at which point
Please cite this article in press as: T. Velegraki, D. Mantzavinos, Solar phreactor, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.0
he wine is ready to be bottled for further maturation or marketingurposes [4]. All the aforementioned processes require the use ofigh volumes of water for washing activities i.e. floor washing from
accidental spills of grape juice and/or wine, equipment cleaning, aswell as fermentation tank and bottle rinsing; a rough estimate isthat for each liter of wine produced, about 1.5 L of wastewater isgenerated alongside.
The main organic content of winery wastewater (WWW) com-prises of soluble sugars (fructose and glucose), various organic acids(tartaric, lactic and acetic), alcohols (glycerol and ethanol) and high-molecular-weight compounds, such as esters, polyphenols, tanninsand lignin [5,6].
The presence of inorganic ions (i.e. potassium and sodium, withlow levels of calcium and magnesium) is mainly owed to the useof cleaning agents, stabilizers and/or pesticide residues [7–9]. Theprecise composition, however, is extremely difficult to assess, asWWW is subject to seasonal variations in both volume and quality(e.g. vintage and non-vintage periods) while also adopts its spe-cific characteristics due to differences in vinification processes andtechniques, grape varietal and amounts of water that each wineryuses; in this context, COD, BOD5 and pH values have been reportedto range from 320 to 296,000 mg L−1, 125 to 130,000 mg L−1 and 3to 12, respectively [5,8,10–14].
As the main volume (>90%) of WWW is produced during
oto-Fenton treatment of winery effluents in a pilot photocatalytic6.008
the harvesting period, (i.e. contains a major fraction of highlybiodegradable compounds such as sugars), most studies addressthe issue of WWW treatment by employing biological processeseither as single treatment or integrated with a physicochemical
treatment level can provide an estimate of pure error, the otheradvantage of running center point replicates in the design is inchecking for the presence of curvature which investigates whetherthe model between the response and the factors is linear.
Table 1Quality characteristics of winery wastewater sampled during vinification season.
rocess [15–17]; however, due to the extreme seasonal varia-ions in WWW production (i.e. both in generated volume andhysicochemical characteristics) there are practical limitationso the operation of a biological system (i.e. demands for con-inuous operation with constant inlet flow characteristics); inhis context, conventional strategies often result in inadequatereatment.
Furthermore, WWW present an additional challenge owed tohe presence of polyphenols; a study performed by Vlyssides et al.18] revealed that the polyphenolic content in WWW during pro-uction of white and red wine may be as high as 280 mg L−1 and450 mg L−1, respectively. The problem with polyphenols arisesue to their potential phytotoxicity and proven resistance to aero-ic degradation [19,20].
Greek legislation dictates the use of a treatment method forhe wastewater generated during the production of wine. As mostineries in Greece are small scale facilities (i.e. less than 2000 t
f wine produced), they are obliged to adopt a septic tank leacheld as a treatment technique, which is, nonetheless, a very basicreatment option with serious drawbacks; inadequate treatmenteads to surrounding soil pH modifications, but also to accu-
ulation of ions (e.g. Na+, Ca2+,NO3−) and/or metals (e.g. Zn+,
u2+, Fe2+), thus posing a risk for plant growth inhibition andhe occurrence of modifications in soil microbial communities21,22].
The aforementioned limitations further stress the need to adopteactive systems much more effective and whose operation woulde unaffected by the dynamic nature of WWW [5,23,24].
Advanced Oxidation Processes (AOPs) constitute a special classf oxidation techniques which usually operate at or near ambientemperature and pressure, exploiting the high reactivity and non-electivity of hydroxyl radicals [25,26]. Photo-Fenton process haseceived much attention for the treatment of industrial effluentsnd elimination of organic pollutants, as it is considered the mostpt of all AOPs to be driven by sunlight; the formation of solubleron-hydroxy and iron-organic acid complexes extents the adsorp-ion wavelength toward the visible region (400 nm < � < 700 nm)o that sunlight can be more efficiently exploited for the detoxi-cation of heavily polluted effluents with significantly lower cost27]. The main drawback of photo-Fenton process is the necessityo work at acidic pH values (i.e. around 2.8) in order to keep theron in soluble form, thus maintaining high degradation efficacy;onetheless, this should not pose a major problem for WWW asheir inherent pH value is, according to literature, less than 5.5 andn many cases lies between 3 and 4 [28]. The low pH values of WWWre ascribed to the presence and/or formation of low moleculareight organic acids (e.g. acetic acid, citric acid and tartaric acid)
29].So far, there are numerous studies dealing with treatment of
inery effluents with photocatalysis, however, only one study hasssessed the treatability of actual wastewater [30], whereas theajority has used simulated winery effluents by diluting wine or
rape juice with deionized water [31–34].The present work aims at assessing a homogeneous solar-driven
enton process (hv/Fe2+/H2O2) for the elimination of the organicontent of actual winery wastewater. The effect of photo-Fentoneagents’ initial concentration has been assessed with prelimi-ary experimental runs in a lab-scale solar simulator by employingxperimental design methodology; the influence of iron (Fe2+) andydrogen peroxide (H2O2) initial concentrations has been eval-ated and further optimization in the pilot-scale photocatalyticeactor has been performed. The main goals of the pilot scale study
Please cite this article in press as: T. Velegraki, D. Mantzavinos, Solar phreactor, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.0
ere to monitor the mineralization of the effluent under limitingnd excess oxidant conditions, to assess the degradation kineticst different iron concentrations and to evaluate the toxicity of thereated effluent toward Aliivibrio fischeri.
PRESSsis Today xxx (2014) xxx–xxx
2. Experimental
2.1. Materials and methods
2.1.1. ChemicalsSolar Fenton experiments were performed using iron sulphate
(FeSO4·7H2O, Riedel-de Haen), reagent-grade hydrogen perox-ide (30%, w/w, Merck) and H2SO4 for pH adjustment (95–97%,Merck). The residual hydrogen peroxide was removed from thetreated samples with catalase (Micrococcus lysodeikticus, Fluka Bio-chemika).
2.1.2. Winery wastewaterThe effluent was collected from a local winery at Chania, W.
Crete, Greece (200 t yr−1 wine production) shortly after the vini-fication period (January–February) when wine stabilization andfiltration processes were completed. The effluent had undergonepreliminary physical treatment (i.e. primary sedimentation) andcontained high amounts of wine vinasse. Frequent measurementsshowed that it was chemically and biologically stable throughoutthe time of experimentation.
All samples were analyzed before use for a number of qual-ity characteristics, which are summarized in Table 1. These valueswere obtained from multiple sample analyses (at least triplicatefor each sample) and are the average values of the parametersmeasured. All parameters were measured according to standardmethods [35]. It is worth noting that the WWW collected containshigh amounts of acetic acid which is an intermediate compoundgenerated along with the degradation of the long-chain acidspresent in wine such as tartaric, malic and lactic acids.
2.1.3. Experimental designThe evaluation of the effect of the Fenton reagent concentra-
tions (i.e. iron and hydrogen peroxide initial concentrations) inthe photocatalytic degradation of the raw winery effluent (i.e.CODo = 1200 ± 150 mg O2 L−1) (Table 1) was based on an experi-mental design approach; a full experimental design, consisting of9 experiments – including the central points for statistical consis-tency – was composed, to assess the effect of the two independentvariables assuming two values or levels (i.e. low and high, indicatedby −1 and +1 coded values, respectively). The low level correspondsto 5 mg L−1 iron and 100 mg L−1 hydrogen peroxide, while the highlevel corresponds to 25 mg L−1 and 900 mg L−1, respectively.
Running multiple replicates at the center point provides an esti-mate of pure error. Although running multiple replicates at any
oto-Fenton treatment of winery effluents in a pilot photocatalytic6.008
Table 222 Experimental design with five replicates at the center point.
Experimentalseries
Actual levels of variables Response factor (Y)
Hydrogen peroxide Iron Residual COD (%)a
22 Experimentaldesign
100 25 86.7900 5 78.1100 5 81.0900 25 62.0
Center runs
500 15 76.6500 15 71.2500 15 71.0500 15 75.7
ttwsaodlT
solocTrrta
wpaHwStc
2
tralooauetopow
t
tocatalytic treatment. The inhibition of the bacteria exposed to
500 15 73.8
a After 3 h treatment.
The variables taken into consideration for the study were the ini-ial concentrations of hydrogen peroxide and dissolved iron whilehe remaining COD (% of its initial value) after 3 h of irradiationas considered as the response factor. The design consisted of two
eries of experiments: (i) a 2k full factorial design (where k = 2 vari-bles), resulting in four experiments with all possible combinationsf the coded variables, (ii) five replicates at the center point of theesign (coded value 0). The experimental design matrix, natural
evels, variable ranges and response factor values are shown inable 2.
The optimization study was performed in a bench-scale solarimulator (Oriel, model 96000) equipped with a 150 W xenon,zone-free lamp and an Air Mass 1.5 Global Filter (cut off of wave-engths <280 nm), simulating solar radiation reaching the surfacef the earth. All experimental runs were performed in a photo-hemical double wall batch reactor made of borosilicate glass.he incident photon flux in the photochemical reactor in the UVegion amounts up to 58 × 10−8 E (L s)−1 (measured actinomet-ically using 2-nitrobenzaldehyde as the chemical actinometer);his corresponds to an irradiance of 7.5 W m−2 which is computedccording to the procedures described in Ref. [36].
In a typical experimental run 0.3 L of pre-filtered (30 �m, PALL)inery wastewater were loaded in the reaction vessel. An appro-riate amount of FeSO4·7H2O solution was inserted immediatelyfter pH adjustment (i.e. pH = 2.8). A pre-determined volume of 30%2O2 (Merck) was inserted in the reaction medium and the lampas turned on. This was considered the time zero of the reaction.
amples were withdrawn at frequent time intervals and were fil-ered (0.45 �m, RC, PALL) prior to each analysis. When necessary,atalase was used to eliminate the residual hydrogen peroxide.
.1.4. Pilot plant reactorA pilot compound parabolic collector (CPC) was used to treat
he winery effluents by the solar light-driven Fenton process. Theeactor comprises glass tubes, through which the effluent flows in
meandering motion, mounted on a fixed platform tilted at theocal latitude (35◦) and it is capable of operating in either batchr continuous mode. The overall capacity of the reactor consistsf the total irradiated volume and the dead volume (tank, pipingnd valves). The maximum capacity of the pilot plant can reachp to 200 L, however, in order to minimize the dead volume effect,xperiments were performed with 105 L of effluent. The intensity ofhe solar UV irradiation was measured with a UV sensor mountedn a Davis weather station located next to the pilot plant whichrovides data in terms of incident W m−2. With Eq. (1), combinationf the data from several days’ experiments and their comparisonith other photocatalytic experiments is possible [26]:
Please cite this article in press as: T. Velegraki, D. Mantzavinos, Solar phreactor, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.0
30 W,n = t30 W,n−1 + �tnUV
30Vi
VT; �tn = tn − tn−1 (1)
PRESSsis Today xxx (2014) xxx–xxx 3
where tn is the experimental time for each sample, UV is the aver-age solar ultraviolet radiation measured during �tn, and t30 ,W is a“normalized illumination time”. In this case, time refers to a nor-malized solar UV power of 30 W m−2 (typical solar UV power ona perfectly sunny day around noon). VT is the total volume of thewater loaded in the pilot plant (105 L) and Vi is the total irradiatedvolume (70 L).
Initially, the reservoir tank was filled up to 105 L with efflu-ent and the air blower was turned on. An initial sample wastaken and a pre-measured volume of concentrated H2SO4 95–97%was introduced for pH adjustment at 2.9 ± 0.07. Following, a pre-determined amount of FeSO4·7H2O was added and the mixturewas homogenized for 15 min to achieve complete dissolution of theiron. An additional sample was withdrawn and analyzed to ensurethat the desired dissolved iron concentration was attained. At thatpoint the appropriate volume of hydrogen peroxide was introducedinto the reactor and the feed pump was turned on to commence thecirculation of the reaction mixture. The solar parabolic collectorswere initially completely covered to avoid any interference fromsolar irradiation. After 30 min, a sample was taken to determine anydegradation due to Fenton reactions and the collectors were uncov-ered. This was considered as zero-illumination time of the photoFenton process. Samples withdrawn from the reactor at frequenttime intervals were conditioned with catalase – once sampling forH2O2 determination had been carried out – in order to eliminateresidual oxidant concentration and avoid further reaction insidethe sampling vial. All samples were analyzed within the same dayof the experimental run.
2.1.5. Analytical determinationsThe extent of mineralization of the organic content was quan-
tified by means of dissolved organic carbon (DOC) and chemicaloxygen demand (COD). COD was determined colorimetrically usinga DR/2010 spectrophotometer (Hach Company, USA) accordingto the EPA approved reactor digestion method, while DOC mea-surements were carried out by injection of filtered samples (RC0.45 mm, PALL) into a Shimadzu-5050A TOC analyzer. Each sam-ple prior to DOC analysis was acidified with H2SO4 solution 1 Nand was subjected to air purging for 10 min to eliminate theinorganic carbon content of the effluent. pH was determinedwith a Crison GLP 21 pH meter. The residual concentration ofhydrogen peroxide was determined with the photometric deter-mination of a yellow-orange titanium oxysulfate complex whichis formed during the reaction of hydrogen peroxide with titaniumoxysulfate solution, with maximum absorbance at 410 nm. A cali-bration curve of high linearity (R2 = 0.99) was developed employingstandard hydrogen peroxide solutions with concentrations rangingfrom 0 to 840 mg L−1. Dissolved iron concentration was mea-sured using the 1,10-phenanthroline spectrophotometric method(ISO 6332:1982). Absorbance at 410 nm (hydrogen peroxide) and510 nm (dissolved iron) was measured in a UNICAM Helios spec-trophotometer. All samples were filtered through 0.45 �m Nylonmembrane filters prior to analysis. Following sampling, residualH2O2 was removed by adding appropriate amounts of catalase solu-tion (17,000 U mL−1), after sample pH adjustment between 6 and8.
The luminescent marine bacteria A. fischeri was used to assessthe acute ecotoxicity of winery wastewater samples during pho-
oto-Fenton treatment of winery effluents in a pilot photocatalytic6.008
treated samples for 15 min was measured using a Microtox 500Analyzer (SDI, USA) according to the Microtox 82% screening test[37]. Each sample was run in duplicate.
The photo-Fenton parameters (i.e. iron and hydrogen perox-de initial concentrations) were evaluated with statistical analysissing the software MINITAB®. A factorial fit was performed for theata presented in Table 2. The analysis of variance (ANOVA) gives aummary of the main effects and the interactions’ effect indicatingheir statistical significance (i.e. p-values of less than 0.05 indicate
significant effect) and it displays both the sequential sums ofquares (Seq SS) and adjusted sums of squares (Adj SS) (Table 3). Asan be seen from Table 3, the main effects and the two-way interac-ion are statistically significant (i.e. p < 0.05), whereas no curvatureas detected (p = 0.126 > 0.05), which implies that a linear model
est describes the experimental system.In turn, the effect of each individual term was assessed by
mploying the Pareto plot (Fig. 1) which allows the visual iden-ification of the statistically important effects of each term andompares their relative magnitude by displaying their absolute val-es. As can be seen in Fig. 1, there are basically two effects whichre statistically important for COD oxidation (i.e. exceeding the ref-rence line and their p-values < 0.05), namely, in decreasing orderf significance: the hydrogen peroxide initial concentration andhe interaction between hydrogen peroxide and iron initial con-entrations. Both significant effects are negative indicating that anncrease in their level promotes COD removal (i.e. decreases theesidual COD).
Even though the iron concentration seems statistically insignif-cant (p = 0.111 > 0.05), thus any variation caused from this variable
ay be explained as random noise, nonetheless it cannot be ruledut during the development of a mathematical model, since thenteraction term was found statistically significant; therefore, theres need to further investigate the interaction effect as there may be
Please cite this article in press as: T. Velegraki, D. Mantzavinos, Solar phreactor, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.0
masking effect of the main effect of iron concentration.Fig. 2(a) and (b) displays the main effects and interaction effect
lots, representing the effect of each variable on the response
ig. 1. Pareto plot of the standardized effects on the residual COD (%) after 3 hreatment.
Fig. 2. Main effects plot (a) and interaction plot (b).
factor. This type of representation shows how the variation of eachparameter affects the response factor and to what extent (Fig. 2(a)),but it also shows the impact of changing the settings of one factorhas on another factor (Fig. 2(b)).
Fig. 2(a) shows that hydrogen peroxide has a higher main effectthat iron; that is, the line connecting the mean responses for dif-ferent concentrations of H2O2 has a steeper slope than the lineconnecting the mean responses at the low and high settings ofiron concentration. Although the amount of iron appears to havea low effect on the response, it is very important to look at theinteraction. From Fig. 2(b) it is evident that the main effect of theiron initial concentration is “masked” by the interaction; the dif-ference in direction of the effects means that at high oxidant levels(i.e. 900 mg L−1) an increase in iron concentration promotes theresponse factor, whereas at low oxidant levels (i.e. 100 mg L−1)the same increase in iron concentration inhibits the response. Inaddition, the interaction plot shows that there is a difference ofmagnitude amongst the two effects which means that the effectof iron concentration is less pronounced when hydrogen peroxideconcentration is at the low level i.e. 100 mg L−1. In this context,neither variable can be considered independently, therefore bothvariables are considered significant. The following model was pro-duced:
Y = 75.122 − 6.9[H2O2] − 2.6[Fe] − 5.4[H2O2][Fe] (2)
with variables in coded levels and Y the residual COD (%) after 3 hof treatment.
oto-Fenton treatment of winery effluents in a pilot photocatalytic6.008
The model presented a high determination coefficient R2 = 87%,explaining 87% of variability in the response. The adjusted deter-mination coefficient (R2
adj = 79.3%) is also high, indicating a highsignificance of the model [38].
ig. 3. Surface plot of residual COD (%) vs hydrogen peroxide and iron initial con-entrations. CODo = 1200 ± 150 mg O2 L−1, pHo = 2.8.
Fig. 3 is a three-dimensional representation of the effect of thenitial concentrations of H2O2 and iron on residual COD (%) fol-owing 3 h of treatment. At low iron concentrations (i.e. 5 mg L−1)he COD abatement enhancement with increasing initial hydrogeneroxide concentration from 100 to 900 mg L−1 is so marginal (i.e.rom 19% to 21.9%) that it can be asserted that no actual improve-
ent is observed due to ‘scavenging’ reactions that ultimately limithe process efficacy. On the other hand, at high iron concentrationsi.e. 25 mg L−1) the beneficial effect of increasing amounts of hydro-en peroxide is more pronounced owed to the generation of highermounts of HO•.
.2. Pilot scale blank experiments
Preliminary experiments were performed to assess the effect ofolar irradiation, with and without the addition of 5 mg L−1 Fe, asell as the effect of dark Fenton reactions occurring under the influ-
nce of 5 mg L−1 Fe and 500 mg L−1 H2O2 at pH = 2.8. The hydrogeneroxide initial concentration was selected at 500 mg L−1 in ordero ensure good homogenization of the oxidant molecules inside the05 L of bulk effluent and at the same time to avoid any localizedcavenging effects. The solar light had similar influence in the CODeduction either in the presence or in the absence of iron catalyst,eading in both cases to less than 5% COD removal after 188 min ofolar irradiation at constant intensity 30 W m−2 (data not shown).
Please cite this article in press as: T. Velegraki, D. Mantzavinos, Solar phreactor, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.0
imilarly, dark Fenton brought on about 10% COD removal for theame time period (data not shown), which is indicative of the lownfluence of dark Fenton reactions to the mineralization of the raw
inery wastewater.
ig. 4. Solar Fenton mineralization of raw winery effluent with hydrogen peroxide in excmount of H2O2 consumed during the reaction and refer to the right Y-axis.
PRESSsis Today xxx (2014) xxx–xxx 5
3.3. Pilot scale solar Fenton treatment of raw winery wastewaters
In order to assess the efficacy of the solar Fenton process on themineralization of the winery effluent from the vinification periodnumerous experimental runs were carried out at pHo = 2.8 undervarying experimental conditions.
Fig. 4 shows the COD abatement of the effluent achieved using5 mg L−1 Fe2+ and 500 mg L−1 H2O2 as initial oxidant concentra-tion (i) without further addition of oxidant throughout the run (i.e.H2O2 limiting) and (ii) with further additions of oxidant duringthe reaction so hydrogen peroxide is always available (i.e. H2O2 inexcess). It can be seen that with limiting H2O2 the COD reduction isapproximately 30–35% after ca. 188 min of normalized illuminationtime, thus revealing the beneficial role of solar light-induced reac-tions compared to the corresponding dark Fenton process (ca. 10%COD abatement, data not shown). With the complete consump-tion of the initially added 500 mg L−1 H2O2 the mineralization ofthe effluent reaches 49% after 280 min of irradiation time. Inter-estingly, a similar removal percentage (i.e. 47%) is reached duringthe run with H2O2 in excess once 500 mg L−1 of oxidant has beenconsumed; however, this is achieved somewhat earlier in the pro-cess i.e. after 200 min, which indicates higher reaction rate and canbe explained by the higher availability of oxidant molecules in thebulk liquid (as the hydrogen peroxide concentration is maintainedabove 100 mg L−1 at all times) and more rapid utilization of theformed oxidant species.
Extending the latter reaction up to 400 min of constant illu-mination, the COD abatement exceeds 80% with ca. 1320 mg L−1
hydrogen peroxide being consumed up to that point. The slightlylower reaction rate observed during the final stages of the exper-imental run (i.e. over 312 min) could be due to the decreasingoxidant concentration in the bulk liquid (less than 100 mg L−1)(i.e. lower availability of H2O2 molecules) as no further additionof hydrogen peroxide was attempted beyond that point.
Bearing in mind that the occurrence of adequate amounts ofoxidant during the solar Fenton process is a detrimental factor inregards to the continuation of the reactions and the mineralizationof the effluent, it was decided to carry out the following experi-ments under conditions of excess hydrogen peroxide, maintainingits concentration between 100 and 500 mg L−1 at all times, thusavoiding the occurrence of recombination and scavenging reac-tions.
The influence of initial iron concentration (e.g. 5 mg L−1 Fe and
oto-Fenton treatment of winery effluents in a pilot photocatalytic6.008
25 mg L ) in the mineralization of raw winery wastewater withsolar Fenton treatment under conditions of excess hydrogen per-oxide (i.e. initial H2O2 500 mg L−1 and stepwise additions that
ess (�) and in limiting conditions (�). The respective hollow symbols represent the
ARTICLE IN PRESSG ModelCATTOD-9114; No. of Pages 7
6 T. Velegraki, D. Mantzavinos / Catalysis Today xxx (2014) xxx–xxx
F initiah efer to
mF
icm2derplo
te
easinisrt
ig. 5. Solar Fenton mineralization of raw winery effluent under excess H2O2 andollow symbols represent the amount of H2O2 consumed during the reaction and r
aintain oxidant concentration 100–500 mg L−1) is presented inig. 5.
It can be observed that the mineralization degree (i.e. 80%)s achieved under both runs, irrespective of the initial iron con-entration used. However, the irradiation time in which thisineralization is reached varies from 220 min to 340 min for
5 mg L−1 and 5 mg L−1 Fe, respectively. By increasing the catalystose, higher reaction rates are observed which can be attributed tonhanced H2O2 decomposition which in turn increases hydroxyladical production. This is corroborated with the lower hydrogeneroxide consumption (i.e. 1270 mg L−1) that is measured when
ow catalyst dose is used, signifying a more effective use of thexidant.
In addition, the presence of high iron concentration underlineshe need for an additional treatment step that will remove thexcess iron from the effluent stream prior to its final discharge.
The gradual decrease in the organic content of the raw win-ry effluent treated with solar Fenton process under excess H2O2nd 5 mg L−1 iron, along with the evolution in eco-toxicity are pre-ented in Fig. 6. The COD abatement commences even from thenitial stages of the treatment and reaches ca. 80% after 402 min oformalized illumination time. The same mineralization percentage
Please cite this article in press as: T. Velegraki, D. Mantzavinos, Solar phreactor, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.0
s achieved in regards to the DOC content of the effluent after theame period of time, however it can be seen that during the initialeaction times no mineralization occurs; as the reaction proceeds,he CO2 formation commences thus decreasing gradually the DOC
Fig. 6. COD (dark gray bars) and DOC (light gray bars) abatement of raw winery efflue
l concentrations of iron catalyst at (�) 5 mg L−1 and (�) 25 mg L−1. The respective the right Y-axis.
content of the effluent. The DOC abatement is less pronounced thanthe respective COD abatement which is attributed to the formationof intermediate organic compounds formed through partial oxida-tion reactions; even so, high DOC removal (i.e. 80%) is observed atthe final treatment stages.
In an attempt to examine the applicability of a solar-inducedphoto-Fenton process for WWW treatment, a-back-of-the-envelope calculation of the solar collector’s surface was performed;this was based on the actual amount of accumulated UV energyrequired to obtain over 80% mineralization of the WWW used inthe present study (i.e. 63 kJ L−1) and the assumption that the annualsolar average UV irradiation in Crete is 20 W m−2 (on a 12-h-per-day basis and 80% availability factor, the number of effective hoursper year is 3504). In this context, to treat 10 m3 of WWW (i.e. thevolume produced annually in a winery of similar capacity and prac-tices during the vinification period) it would require an irradiatedsurface of 2.5 m2. In case the WWW generated during harvest-ing period is also included, the treated volume would rise up to300 m3 and the corresponding irradiated area would increase to75 m2. The major operational costs of such a plant would reflecton the consumption of reagents and more specifically on hydro-gen peroxide. Bearing in mind that technical grade H2O2 (30%)costs no more than about 1 D kg−1 [39,40] and 1.27 kg H O m−3 is
oto-Fenton treatment of winery effluents in a pilot photocatalytic6.008
2 2required for over 80% mineralization, this would result in approxi-mately 4.2 D m−3 for wastewater treatment; this adds up to about1260 D yr−1 considering an annual WWW production of about
nt and toxicity evolution (x) during solar Fenton oxidation in the solar reactor.
[[39] R. Bauer, H. Fallmann, Res. Chem. Intermed. 23 (1997) 341–354.
ARTICLEATTOD-9114; No. of Pages 7
T. Velegraki, D. Mantzavinos /
00 m3 for a medium-sized winery (like the one that suppliedhe samples in this work). In addition, the amount of ferrous salteeded would not exceed 40 kg yr−1 (i.e. to treat 300 m3 at the max-
mum concentration of 25 mg L−1 Fe tested in this work) at a cost of.5–1 D kg−1 [39,40]. With a very rough estimation that a wineryf this capacity makes about 200,000 D yr−1 in turnover, we canafely assume that the treatment cost does not put at risk the eco-omic viability of the winery (i.e. costs associated with the Fenton’seagents are well below 1% of winery turnover).
The raw effluent shows moderate toxicity (e.g. ca. 40% inhi-ition) prior to oxidation (Fig. 6), which is below the 50%ioluminescence bacteria inhibition threshold; as the solar Fentoneaction commences, the bacteria inhibition increases up to almost0% which indicates that some intermediate compounds formeduring the first stage of oxidation are potentially toxic to the marineacteria A. fischeri [41]. Proceeding at higher reaction times withver 50% COD abatement, the toxicity decreases significantly withhe inhibition reaching down to 20% and is maintained at such lowevels even after the COD of the effluent has been reduced over0%; this is probably due to the exposure of the bacteria to a signif-
cantly more complex matrix – such as the actual winery effluentsed presently – than their inherent saline environment.
. Conclusions
The conclusions drawn from the present study can be summa-ized as follows:
The photo-Fenton process utilizing solar energy proved to beighly efficient in the mineralization and detoxification of real win-ry wastewater produced during the vinification period. Organicatter degradation reached COD and DOC removal values as high
s 80%, after 402 min (t30 ,W) of treatment, which correspond to3 kJ L−1 of accumulated UV energy. The resulting effluent exhib-
ted very low toxicity using A. fischeri test as toxicity bioassay. Theegree of mineralization is irrespective of the initial iron concen-ration used; however, with increasing catalyst dose the reactionate is enhanced due to higher hydroxyl radical production. Themployment of low (i.e. 5 mg L−1) rather than high (i.e. 25 mg L−1)ron concentration signifies a more effective use of the oxidantn order to reach similar mineralization. At low catalyst dose (i.e.Fe2+] = 5 mg L−1) mineralization (i.e. ca. 50%) is dependent on thexidant consumption (i.e. 500 mg L−1), irrespective of the excessxidant present; however, shorter reaction times are requirednder excess H2O2 due to higher availability of oxidant molecules
n the bulk effluent.
cknowledgments
This study was performed, with the contribution of the LIFEnancial instrument of the European Union, as part of the LIFEroject “Advanced systems for the enhancement of the envi-onmental performance of WINEries in Cyprus” WINEC LIFE08
Please cite this article in press as: T. Velegraki, D. Mantzavinos, Solar phreactor, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.0
NV/CY/000455.The authors would like to acknowledge the help received
t Stylianoudakis Winery for the collection and delivery of theastewater.
[
[
PRESSsis Today xxx (2014) xxx–xxx 7
References
[1] http://www.wineinstitute.org/files/2010 World Wine Production by Country.pdf.
[2] S.A. Ene, C. Teodosiu, B. Robu, I. Volf, J. Clean. Prod. (2013) 122–135.[3] R.S. Marshall, M. Cordano, M. Silverman, Bus. Strategy Environ. (2005) 92–109.[4] http://en.wikipedia.org/wiki/Winemaking#Harvesting and destemming.[5] K.P.M. Mosse, A.F. Patti, E.W. Christen, T.R. Cavagnaro, J. Aust, Grape Wine Res.
17 (2011) 111–122.[6] L. Malandra, G. Wolfaardt, A. Zietsman, M. Viljoen-Bloom, Water Res. 37 (2003)
4125–4134.[7] J.A. Chapman, R.L. Correll, J.N. Ladd, J. Aust, Grape Wine Res. 1 (1995) 39–47.[8] M.A. Bustamante, C. Paredes, R. Moral, J. Moreno-Caselles, A. Perez-Espinosa,
M.D. Perez-Murcia, Water Sci. Technol. 51 (2005) 145–151.[9] C. Sheridan, D. Glasser, D. Hildebrandt, J. Petersen, J. Rohwer, S. Afr. J. Enol. Vitic.
32 (2011) 1–8.10] R. Mosteo, J. Sarasa, M.P. Ormad, J.L. Ovelleiro, J. Agric. Food Chem. 56 (2008)
7333–7338.11] A. Eusebio, M. Petruccioli, M. Lageiro, F. Federici, J. Duarte, J. Ind. Microbiol.
Biotechnol. 31 (2004) 29–34.12] P. Artiga, E. Ficara, F. Malpei, F. Garrido, R. Mendez, Desalination 179 (2005)
161–169.13] M. Petruccioli, J. Duarte, F. Federici, J. Biosci. Bioeng. 90 (2000) 381–386.14] W.C. Quayle, A. Fattore, R. Zandona, E.W. Christen, M. Arienzo, Water Sci. Tech-
nol. 60 (2009) 2521–2528.15] C. Valderrama, G. Ribera, N. Bahí, M. Rovira, T. Giménez, R. Nomen, S. Lluch, M.
Yuste, X. Martinez-Lladó, Desalination 306 (2012) 1–7.16] L. Serrano, D. de la Varga, I. Ruiz, M. Soto, Ecol. Eng. 37 (2011) 744–753.17] L.A. Ioannou, D. Fatta-Kassinos, J. Environ. Chem. Eng. 1 (2013) 703–712.18] A.G. Vlyssides, E.M. Barampouti, S. Mai, Water Sci. Technol. 51 (2005) 53–60.19] M. Arienzo, E.W. Christen, W. Quayle, N. Di Stefano, Water Environ. Res. 81
Water Manage. 120 (2013) 5–10.23] M.S. Lucas, J.A. Peres, G. Li Puma, Sep. Purif. Technol. 72 (2010) 235–241.24] R. Ganesh, R. Rajinikanth, J. Thanikal, R. Ramanujam, M. Torrijos, Bioproc.
Biosyst. Eng. 33 (2010) 619–628.25] R. Andreozzi, C. Caprio, A. Insola, R. Marotta, Catal. Today 53 (1999) 51–59.26] S. Malato, J. Blanco, A. Vidal, D. Alarcon, M.I. Maldonado, J. Caceres, W. Gernjak,
Sol. Energy 75 (2003) 329–336.27] S. Malato, P. Fernandez-Ibanez, M.I. Maldonado, J. Blanco, W. Gernjak, Catal.
Today 147 (2009) 1–59.28] G. Levay, Effluent Management for Wineries and Distilleries. The Australian
Regulatory Framework and Code of Practice, Ian Wark Research Institute, Uni-versity of South Australia, Adelaide, 1995.
29] L.H. Van Schoor, Omgewingswetgewing in die wingerd en wynbedryf, WynboerTegnies 138 (2001) 114–117.
30] J.M. Monteagudo, A. Duran, J.M. Corral, A. Carnicer, J.M. Frades, M.A. Alonso,Chem Eng. J. 181–182 (2012) 281–288.
31] M.P. Ormad, R. Mosteo, C. Ibarz, J.L. Ovelleiro, Appl. Catal., B: Environ. 66 (2006)58–63.
32] P. Navarro, J. Sarasa, D. Sierra, S. Esteban, J.L. Ovelleiro, Water Sci. Technol. 51(2005) 113–120.
33] R. Mosteo, P. Ormad, E. Mozas, J. Sarasa, J.L. Ovelleiro, Water. Res. 40 (2006)1561–1568.
34] M.S. Lucas, R. Mosteo, M.I. Maldonado, S. Malato, J.A. Peres, J. Agric. Food Chem.57 (2009) 11242–11248.
35] American Public Health Association, Standard Methods for the Examination ofWater and Wastewater, 20th ed., American Public Health Association, Wash-ington, DC, 1978.
