Journal of Environmental Management 201 (2017) 177e189

Contents lists available at ScienceDirect

Journal of Environmental Management

journal homepage: www.elsevier.com/locate/jenvman

Research article

Evaluation of the treatability of a winery distillery (vinasse)

wastewater by UASB, anoxic-aerobic UF-MBR and chemical
precipitation/adsorption
Luigi Petta a, Sabino De Gisi b, *, Patrizia Casella c, Roberto Farina a, Michele Notarnicola b
a
ENEA, Department for Sustainability of Production and Territorial Systems, “Water, Waste and Raw Materials Integrated Management Technologies” Lab.,
Via Martiri di Monte Sole n. 4, 40129, Bologna, BO, Italy
b
Department of Civil, Environmental, Land, Building Engineering and Chemistry (DICATECh), Polytechnic University of Bari, Via E. Orabona n. 4, 70125, Bari,
BA, Italy
c
ENEA, Department for Sustainability of Production and Territorial Systems, “Environmental Biogeochemistry” Lab., Piazzale Enrico Fermi n. 1, 80055,
Portici, NA, Italy

a r t i c l e i n f o a b s t r a c t

Article history: A multi-stage pilot-scale treatment cycle consisting of an Upflow Anaerobic Sludge Blanket reactor
Received 8 January 2017 (UASB) followed by an anoxic-aerobic Ultra Filtration Membrane Bio Reactor (UF-MBR) and a post
Received in revised form treatment based on chemical precipitation with lime or adsorption on Granular Activated Carbons (GAC),
12 June 2017
was applied in order to evaluate the treatment feasibility of a real winery distillery wastewater at lab-
Accepted 18 June 2017
oratory and bench scale. The wastewater was classified as high strength with acidic pH (3.8), and con-
centrations of 44,600, 254, 604 and 660 mg/l for CODtot, total nitrogen, total phosphorous and phenols,
respectively. The UASB reactor was operated at Organic Loading Rates (OLR) in the range 3.0e11.5
Keywords:
Anoxic-aerobic UF-MBR
kgCODtot/m3/d achieving treatment efficiency up to 97%, with an observed methane production of 340 L
Chemical precipitation of CH4/kgCOD. The MBR system was operated with an organic load in the range 0.070e0.185 kgCOD/
Lime kgVSS/d, achieving a removal up to 48%, 67% and 65% of the influent COD, total nitrogen and phenols,
Nitrogen removal respectively. The combination of UASB and UF-MBR treatment units was not effective in phosphate and
UASB colour removal assigning to further chemical precipitation and adsorption processes, respectively, their
Winery distillery wastewater complete removal in order to comply with legal standards for wastewater discharge. Subsequently, the
optimization of the investigated treatment chain was assessed by applying a chemical precipitation step
upstream and downstream the UASB reactor, and a related treatment unit cost assessment is presented
in view of a further technological scale-up.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction vary over the year demanding that the treatment system must be
versatile to face both the loading regimen and stream fluctuation
Processes for wine production generate organic and inorganic (Bolzonella et al., 2010).
pollution mostly associated with solid wastes and liquid effluents. Winery wastewater contains large amounts of biodegradable
The liquid effluents usually referred as “winery wastewater” are organics in addition to relatively small concentrations of recalci-
mainly originated in washing operations during grape harvesting, trant compounds such as polyphenols, organic acids and sugars and
pressing and first fermentation phases of wine processing relatively low presence of solids and nutrients (Serrano et al., 2011;
(Rodríguez-Chueca et al., 2017; Ioannou et al., 2015; Lucas et al., Braz et al., 2010). According to Andreottola et al. (2005) and Beck
2010; Mulidzi, 2010; Mosteo et al., 2007), as well as a result of et al. (2005), the readily biodegradable COD represents the most
the distillation process applied to wine processing residues (e.g. relevant fraction of total COD with values ranging between 71.4%
vinasses). As a consequence, volumes and pollution loads greatly and 85%, respectively. The high percentage of this fraction is due to
the prevalent presence of ethanol and, to a smaller extent, sugars
and organic acids (Andreottola et al., 2009). The concentration of
* Corresponding author. slowly biodegradable COD varies from 2.9% to 9.4% of total COD
E-mail address: [email protected] (S. De Gisi). while the un-biodegradable soluble fraction resulted quite different

http://dx.doi.org/10.1016/j.jenvman.2017.06.042
0301-4797/© 2017 Elsevier Ltd. All rights reserved.
178 L. Petta et al. / Journal of Environmental Management 201 (2017) 177e189

in Andreottola et al. (2005) and Beck et al. (2005) probably due to (i) Identification of the optimal operational set-up and related
the different approaches used for COD fractionation. criticalities of the UASB reactor as well as verification of the process
Nowadays, several winery wastewater treatment technologies efficiency with reference to high organic loading rates (OLR) in the
are available and they involve, in most cases, the use of biological range 6.2e11.5 kg COD/m3/d; (ii) Identification of the optimal
processes (Ioannou et al., 2015; Chai et al., 2014; Mulidzi, 2007; operational set-up and related criticalities of the anoxic-aerobic UF-
Thanikal et al., 2007; Brucculeri et al., 2005; Petruccioli et al., MBR system, assessing the individual contributions referable to the
2002) (see Table 1S, Supplementary material). Such processes biological process and the membrane separation process; (iii)
ensure a significant removal of the organic content although the Identification of the main process parameters of chemical precipi-
presence of recalcitrant compounds frequently makes the complete tation with lime and GAC adsorption, intended as post-treatment
winery wastewater treatment impossible (Ioannou et al., 2015). solutions; (iv) Optimization of the treatment cycle by using
A common, simple and relatively low-cost solution may be chemical precipitation with lime upstream and downstream the
represented by the co-treatment of municipal and winery waste- UASB treatment unit.
water in conventional activated sludge processes (Andreottola et al.,
2009; Ferna ndez et al., 2007; Pathe et al., 2002). Possible problems 2. Materials and methods
such as bulking phenomena or decrease of sludge settleability
suggested the proposition of dedicated plant including aerobic/ 2.1. Experimental plan
anoxic processes using suspended biomass (activated sludge,
membrane bioreactors, MBR, sequencing batch reactors, SBR), aer- The experimental trials were carried out on a real winery
obic biofilm systems (conventional rotating biological contactors, wastewater. A total amount of about 2 m3 was collected from a
RBC, innovative fixed bed biofilm reactors, FBBR, or moving bed distillery located in South Italy (Sicily) and, after a preliminary
biofilm reactors, MBBR, SBBR), anaerobic processes using sus- characterization, the inlet wastewater was properly stored into a
pended biomass (conventional anaerobic digesters or anaerobic stirred and refrigerated (2e4
C) tank, also used as feeding unit.
sequencing batch reactors, ASBR), anaerobic biofilm systems using The pilot scale tests involved three treatment steps arranged in
granules (Upflow Anaerobic Sludge Blanket, UASB), hybrid systems series, as following reported (see Fig. 1a): (i) anaerobic treatment;
(Up flow Sludge Blanket Filter, USBF) (Moletta, 2005; Andreottola (ii) anoxic-aerobic treatment; (iii) chemical and physical post-
et al., 1998) and constructed wetlands (Andreottola et al., 2009). treatment. Regarding the first treatment step, an UASB reactor
The only anaerobic processes, many of which summarized in was used with the intent of maximizing the biogas production to be
Table 1S (see Supplementary material), are not able to ensure used for energy purposes. As second treatment step, an anoxic-
compliance with final discharge standards set by the law. Conse- aerobic MBR system equipped with an ultrafiltration membrane
quently, they are generally used as pre-treatments thus requiring was realized and fed with the clarified fraction of the UASB effluent
additional treatments downstream. (digestate), with the main scope to provide a residual COD removal
Membrane technologies, combined with conventional activated as well as a reduction of N and P content. The third step consisted in
sludge-based processes, have been extensively studied over the two separate treatment options: chemical precipitation based on
years. They represent an interesting solution especially for small the use of lime, and GAC adsorption. The main scope of such
companies as highlighted in Bolzonella et al. (2010). treatments was the residual COD, phosphorous and colour removal
Other technological solutions are based on the combination of in order to comply with the effluent standards for wastewater
different processes, even in a single reactor, as already experienced discharge set by the European (Directive 91/271/EEC) and Italian
in Andreottola et al. (2005), Farina et al. (2004), Petropoulos et al. Law (Legislative Decree No. 152/2006).
(2016) and Molina et al. (2007), Akunna and Clark (2000), Basset Each treatment step was operated separately, the effluent of
et al. (2016) and Andreottola et al. (2002) with reference to FBBR, each treatment unit being stored and refrigerated in order to be fed
ASBR, hybrid USBF, GRABR, AnMBR and SBBR, respectively. How- to the downstream one. At first, the UASB process was operated for
ever, organic matter removal is not the only task for winery 4 months, and the effluent was stored in a refrigerated tank during
wastewater treatment, since nitrogen, phosphorous compounds, last 3 operation weeks before being fed to the MBR section; the
and colour also need to be tackled following a multiple objectives MBR was operated for 4 months and the effluent was collected and
approach (Ioannou et al., 2015). stored during last 2 months of operation and then used for further
In this regard, literature shows several examples such as Amaral- chemical treatments. The whole experimental tests were carried
Silva et al. (2016) who integrated ferric coagulation, Fenton reaction out for 10 months and structured according to the time schedule as
and activated sludge for phosphorus removal. De Gisi et al. (2016) shown in Fig. 1b.
highlighted the use of granular activated carbons (GAC) as well as
alternatively low-cost sorbents for colour removal from different 2.2. Inlet wastewater characterization
wastewater. Regarding membrane technologies, Bolzonella et al.
(2010) adopted a MBR system based on microfiltration (MF) for The real wastewater consists of vinasse resulting from the grape
the biological activated sludge phase, while real cases based on the distillation process. Wastewater was firstly characterized by means
use of ultrafiltration MBR (UF-MBR), are rather limited. Sheldon and of chemical-physical analyses, as reported in Table 1, showing an
Erdogan (2016) have recently applied an UF-MBR for soft drink acidic pH and a high organic matter concentration, with an almost
production wastewater, with different characteristics respect to the all soluble COD including a recalcitrant fraction which is, at least in
winery ones. part (e.g. tannic content), responsible of the dark colour as dis-
In this context, with the intent of strengthening the current cussed hereinafter.
knowledge, the article deals with the verification of the treatability
of a real winery distillery (vinasse) wastewater through the appli- 2.3. Reactor set up
cation of a treatment cycle consisting in UASB, anoxic-aerobic UF-
MBR, and post treatment steps (chemical precipitation with lime or The bench scale UASB system consists in a cylindrical reactor
alternatively GAC adsorption), in order to comply with the made of Plexiglas with an approximate volume of 24 L (H ¼ 75 cm,
discharge standards set by the Italian and European regulations. D ¼ 22 cm); a series of manual ball valves in stainless steel have
More specifically, the following sub-goals have been investigated: been inserted on its surface in order to allow (i) the removal of the
 L. Petta et al. / Journal of Environmental Management 201 (2017) 177e189 179

Fig. 1. Experimental treatment scheme highlighting the principal sections (in red and numbered) for material balances purposes (a); Time schedule of the individual processes
investigated (b). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

microbial sludge samples, (ii) the release of the biogas produced in reactor. Agitation was provided by recycling the mixed liquor from
the UASB reactor as well as (iii) the output of the recirculation flow. an intake below the inert support and injecting it upwards from the
The sampling point of the recirculation flow was installed in cor- bottom of the reactor. The produced biogas was measured by an
respondence of the three phase separator. The recirculation flow Elkro gas (BK-P) meter. Finally, the effluent is discharged from the
was realized by means of a peristaltic pump, with the scope of top of the reactor by means of an overflow weir. The entire UASB
increasing the ascent rate of the liquid in the reactor and to pro- reactor was placed inside a thermostatically regulated heating ca-
mote of the microbial bed and influent mixing. Furthermore, both ble that allows operating at a constant temperature of 37
C.
the feed points of the inlet wastewater and of the soda solution for The working volumes of the anoxic and aerobic MBR compart-
pH regulation inside the reactor were realized on the delivery pipe. ments were 0.85 and 1.0 L, respectively. The system, realized in
For this purpose, on the same line, a glass pH probe connected via a Plexiglas, is constituted by two vessels separated by a polyvinyl-
measuring and control system to the piston pump was positioned. chloride septum. The inlet wastewater is fed into the denitrification
When the probe detects a value of pH lower than the pre- compartment, equipped with a mechanical stirring system. The
determined set point, the pump is activated and feeds soda. internal recycle was carried out using a peristaltic pump (Watson
Considering the inlet wastewater, the feeding flow enters from the Marlow 403U/R1) at a constant flow rate of 5.0 ml/min for the
bottom of the reactor so as to allow the wastewater to pass through entire duration of the study. A second peristaltic pump (Watson
the whole thickness of the microbial bed, participating, thereby, to Marlow 401U/D1), controlled by a level sensor in the reactor, was
its mixing. A volumetric pump (Hydra EM24) was used to fill the used for reactor feeding. A hollow fibre membrane module (ZW1,
Zenon, Italy) with characteristics reported in Table 2S (data from
the supplier, see Supplementary material), was immersed into the
Table 1 aerobic compartment. Membrane filtration was carried out using a
Characteristics of the raw winery distillery wastewater. programmable piston pump (Ismatec, Cole-Parmer, USA) whereas
aeration was obtained using a blower with a constant flow at
Parameter Unit Average values St. Dev
approx 80 L/h. The system operated inside a thermostatically
pH e 3.8 ±0.05 regulated heating cable at a constant temperature of 20
C.
Conductivity mS/cm 8.3 ±0.09
Total Solids (TS) g/l 29 ±0.97
The chemical precipitation treatment was performed by means
Total Volatile Solids (TVS) g/l 18.8 ±0.60 of the Jar Test apparatus. Phosphorus removal was evaluated
Total Suspended Solids (TSS) g/l 0.48 ±0.02 through lime-based precipitation. Trial tests consisted in a first step
Volatile Suspended Solids (VSS) g/l 0.39 ±0.01 aimed at defining the most effective dosage by adding increasing
Total COD (CODtot) mg/l 44,600 ±94.5
amounts of chemicals (as to realize concentrations from 2.0 to 8.0 g
Soluble COD (sCOD) mg/l 44,000 ±51.0
Total Nitrogen (TN) mgN/l 254 ±26.0 Ca(OH)2/l) with a constant reaction time of 1 min, a mixing speed
Ammonia Nitrogen (NHþ 4) mgN-NHþ
4 /l 135 ±10.0 rate of 300 rpm and measuring the pollutants concentration (Total
Total Phosphorous (TP) mg/l 604 ±29.01 phosphorous, phosphorus orthophosphate and COD), in the su-
Phosphates (P-PO4) mg/l 446 ±9.5 pernatant. Once identified the optimal dosage with the application
Total phenols mg/l 660 ±11.53
Total Volatile Fatty Acids (VFA) mg/l 3300 ±66.04
of the system optimization methodology described in literature (De
Sulphates (SO24 ) mg/l 44.7 ±1.76 Gisi et al., 2014; De Feo et al., 2013), different reaction times of 1, 5,
Chlorides (Cl) mg/l 128.8 ±6.96 10 and 20 min have been tested, maintaining all the other operating
Colour e Dark brown e conditions as constant. After settling, the samples were centrifuged
180 L. Petta et al. / Journal of Environmental Management 201 (2017) 177e189

and then measured by spectrophotometric techniques after 5, 7 microbial biomass was facilitated. After the first week, the contin-
and 10 d contact time. uous feed was switched starting from a daily OLR of about 3 kg
The GAC adsorption tests consisted in the arrangement of an COD/m3/d, corresponding to a 16-days hydraulic retention time
activated carbon contact bed (84.5 g dry weight) in a Buchner filter, (HRT). The feed flow rate was progressively increased to reach the
which was put in contact with a slow sample flow of the effluent to load of 11.6 kg COD/m3/d, corresponding to a 4 days HRT. The pH
be treated. Samples were then measured by spectrophotometric inside the reactor was maintained in neutral field setting the con-
techniques. trol system to a threshold of 6.7. For the pH adjustment it was first
used soda 0.05 M and then, as to limit the dilution effect, a soda
2.4. Sampling, analyses and operating conditions for the biological solution prepared with the same wastewater. However, the oper-
processes during start-up and regime ating pH values have always been included in the range 7.3e8.0.
The anoxic-aerobic MBR system operation was monitored over
The operation of the UASB reactor was monitored over four four months (as in the time schedule reported in Fig. 1b). A full
months as shown in Fig. 1b. Regularly, at least on weekly basis, characterization (TSS, VSS, COD, TN, N-NH4, N-NO-3 N-NO-2, P-PO4,
samples were taken at feed and treated effluent, while periodically, Cl, SO2
4 , total phenols, colour) was provided at least on weekly
according to specific experimental needs, track studies to monitor basis in MBR influent and effluent samples in order to monitor the
the trends of filtered COD, VFA (Volatile Fatty Acids), TSS (Total overall treatment performance, while specific parameters (TSS,
Suspended Solids), pH and alkalinity within a cycle treatment were sCOD, N-NO-3 N-NO-2) were performed in both anoxic and aerobic
performed. Total Solids (TS), Volatile Solids (VS), Total (TSS) and compartments in order to monitor the progress of nitrification and
Volatile Suspended Solids (VSS), ammonia nitrogen (N-NHþ 4 ), were denitrification processes. The evaluation of ammonia nitrogen
regularly performed for the untreated and treated effluents ac- concentration in both compartments was not provided because of
cording to the Standard Methods (Standard Methods, 1995). the intense colouration of the wastewater and the reduced reaction
Monovalent anions were analysed using a HPIC (Dionex 5000i). volumes, which did not allow the determination either by spec-
Total alkalinity (TA) was measured by titration at pH 3.8. VFA were trophotometry or after distillation of the sample. The colour of the
determined, on filtered 0.45 mm samples, by gas chromatography effluent was monitored using spectrophotometric scans and
using a DANI 8510 GC equipped with a FID detector. Total and compared with the input in order to verify the effectiveness of the
filtered (0.45 mm) COD was analysed by Dr. Lange kit. The sludge membrane filtration treatment. The reactor was started up with
sampled from the reactor was observed by a phase-contrast light aerobic biomass taken from a municipal wastewater treatment
microscopy (Jenalumar A/D contrast light microscope, 1000 plant, with the characteristics reported in Table 3S (see Supple-
magnification) in bright field phase contrast, Nomarski interfer- mentary Material). Firstly, a feeding of 0.043 L/h, corresponding to
ential contrast and brield field. All parameters (COD, Total Nitrogen an organic loading rate (OLR) of 0.185 kg COD/kg VSS, was
and Ammonia, Phenols) were performed with a minimum fre- considered in order to provide a HRT of about 2 days. Subsequently,
quency of one week in order to monitor the performances of the as better specified hereinafter, the observed trend led to the gradual
anaerobic digestion process and the amount of biogas produced in reduction of the influent flow rate with a subsequent reduction of
the reactor as well as its quality, meant in terms of percentage of the organic loading rate (OLR) as visible in Table 4S (see Supple-
methane content. The amount of biogas was daily compared to the mentary Material). Nitrates recirculation flow rate (from the aero-
expected production in turn evaluated with the following rela- bic to the anoxic compartment) was kept on the value of 0.024 L/h.
tionship (Metcalf and Eddy, 2003): During the experimentation, the surplus sludge from the reactor
was not extracted so as to allow the optimal biomass growth up to
Expected biogas ¼ [(CODIN e CODOUT) $qIN$ Y] / (%CH4) (1) the achievement of desired working concentrations (6e8 gTSS/l).
The pH and the dissolved oxygen in both biological compartments
where: were monitored using portable probes; the aeration in the nitri-
fying unit was from time to time adjusted so as to maintain a dis-
 Expected biogas ¼ production of expected biogas, in L; solved oxygen concentration (DO) in the range 5e7 mg/l, whereas
 CODIN, CODOUT ¼ inlet and outlet values of COD, in g/L; in the anoxic compartment the DO was adjusted by changing the
 qIN ¼ volumetric input flow rate, in L; recirculation flow rate so as to maintain a value lower than 0.5 mg/l.
 Y ¼ specific yield (theoretical) of COD conversion into methane, The pH was maintained in the range of 7e9 in both units. To reduce
in L of CH4 per gram of COD, assumed as 350 L CH4/kg COD for fouling, the membrane device was operated alternating cycles of 3
wastewater (Metcalf and Eddy, 2003); and 2 min of filtration and relaxation, respectively (no back wash).
 % CH4 ¼ percentage of methane in the biogas produced. The transmembrane pressure (TMP) was monitored using a digital
gauge (Cole-Parmer, USA). A mechanical stirrer (RZR, Heidolph,
The percentage of methane was controlled by gas- Italy) was used for mixing the anoxic tank. Instead, a homemade
chromatography with a 2e3 weeks frequency in order to assess wet gasmeter was used for biogas production monitoring. Sludge
the specific heat of the generated biogas. The reactor was started withdrawal was performed manually in order to control the SRT at
with anaerobic biomass taken from a real granular type digester approximately 80e100 days.
operating on the processing of fruit waste, whose initial charac- With reference to the chemical precipitation and GAC adsorp-
teristics are reported in Table 3S (see Supplementary Material). On tion, total phosphorus (TP), phosphate, COD and colour were
a total volume of 24 L of the experimental reactor, about one third measured on effluent samples according to the standard methods
was filled with the fresh granular sludge, while the remaining part (Standard Methods, 1995).
with tap water, obtaining, in this way, an average sludge concen-
tration (on the whole reactor volume) of about 15 g/L. Subse- 3. Results and discussion
quently, the system was launched by providing, during the first
week, a feeding for about 7 h per day, so as to realize an OLR of 3.1. UASB process
1.0 kg COD/m3/d, to be considered as very low if compared to the
typical values applied for UASB-type plants (5e15 kg COD/m3/d) The UASB system has been started with a load of about 3.0 kg
(Andreottola et al., 2009). In this way, the acclimatization of the COD/m3/d, gradually increased up to 11.6 kg COD/m3/d. The results
 L. Petta et al. / Journal of Environmental Management 201 (2017) 177e189 181

in terms of percentage removal are reported in Fig. 2a. phenols into methane.
COD removal rate was roughly constant and equal to 95%, as an In overall, the optimal OLR condition has to be considered that
evidence that almost all of the organic compounds in the vinasse corresponding to 4.5 kg COD/m3/d which allows relatively short
wastewater are relatively simple to degrade. In particular, the COD retention times (HRT ¼ 6 days), good treatment performances and
removal was 97.5, 97.2 and 96.0% in correspondence of OLR values low foam appearance in the reactor, as discussed herein. Despite
of 3.0, 4.5 and 11.5 kg COD/m3/d, respectively (Fig. 2a). Such results this, the concentrations of the main parameters in the effluent did
are in line with Andreottola et al. (2009) who highlights COD per- not comply with the effluent standards set by the Italian Law, with
centage removals in the range 80e97% for an UASB process working specific reference to the average effluent concentrations of COD, TN,
with OLR values in the range 5e15 kg COD/m3/d. N-NHþ 4 and total phenols (Table 2). The observed results were in
As a result of the UASB process, the ammonia nitrogen showed line with Andreottola et al. (2009) who highlighted that the UASB
an average increase of about 18e20% (Fig. 2a), which highlights that treatment applied to winery wastewater is not able to reach the
the UASB treatment provided an almost complete removal of the effluent standard limits given by the Italian Law.
organic nitrogen due to the adsorption into the sludge blanket (as The UASB process has consequently to be considered as a pre-
evidenced by the increased total and volatile solids content in the treatment unit for winery wastewater treatment, according to
anaerobic sludge after the experimental trials) coupled to the several literature examples of biological-based pre-treatments of
ammonification process taking place in the reactor. A certain TN agro-food wastewater, such as in Sheldon and Erdogan (2016), De
removal was observed with an average value of 10.6%, to be Gisi et al. (2013) and Buntner et al. (2013) in the cases of soft
considered as quite physiological and due to a secondary ammonia drink industry wastewater, biodiesel production wastewater and
stripping effect favoured by biogas production, especially during dairy wastewater, respectively. Moreover, the UASB effluent was
the experimental runs when operating pH values of 8e8.5 were characterized by an intense dark-orange colour as evidenced by the
measured in the reactor. spectrophotometric scanning relayed to the inlet and outlet UASB
Concerning other parameters, total phenols removal was on wastewater (Fig. 2b). Even this aspect was clearly reported in
average of 75.0% with lower values of about 65% in correspondence literature: Ioannou et al. (2015) highlighted the possibility to use a
of an OLR of 6.2 kg COD/m3/d. Similar results were obtained by chemical post-treatment based on Fenton's oxidation in order to
Hussain et al. (2008) who, working with an OLR in the range of remove colour from a UASB reactor treating winery wastewater.
4.0e4.1 kg COD/m3/d, showed a high percentage of phenol degra- With reference to UASB process performance in terms of biogas
dation (about 90%) as well as the consequent transformation of production, the amount of produced biogas was detected during

Fig. 2. UASB treatment results: (a) Performance in terms of pollutants removal; (b) Spectrophotometric scanning of the UASB feeding and effluent; (c) Comparison between the
measured and expected biogas productions varying the organic load; (d) Biogas production rate and specific methane production during the experimentation.
182 L. Petta et al. / Journal of Environmental Management 201 (2017) 177e189

Table 2
Characteristics of the winery distillery wastewater downstream to UASB reactor.

Organic load [kgCOD/m3/d] Average values and standard deviation of the main parameters

sCOD [mg/l] TSS [g/l] VSS [g/l] TN [mg/l] N-NHþ

4 [mg/l] Total Phenols
[mg/l]

3.0 1100.0 ±55.2 0.42 ±0.03 0.28 ±0.03 220.0 ±56.0 210.0 ±51.4 142.0 ±8.7
4.5 1260.0 ±54.3 0.90 ±0.06 0.78 ±0.05 240.0 ±58.5 150.0 ±38.6 150.0 ±8.7
6.2 1780.0 ±72.0 0.86 ±0.05 0.72 ±0.04 210.0 ±55.1 146.0 ±36.2 240.0 ±13.7
10.5 1720.0 ±67.0 1.29 ±0.07 1.05 ±0.05 252.0 ±64.3 105.0 ±27.5 n.a.a e
11.5 1790.0 ±69.0 1.47 ±0.08 1.12 ±0.04 221.0 ±59.9 160.0 ±41.3 122.0 ±7.1
Discharge limit standardsb 160.0 e 0.08 e e e 15.0 e 15.0 e 0.5 e
a
n.a. ¼ not available.
b
Italian Law (D. Lgs. 152/2006).

the entire experiment in order to evaluate methane specific pro- production according to the expected values.
duction (in terms of litres of methane produced per kg of COD The characteristics of the biological sludge after 55 days of
removed by the system) and the biogas production rate (in terms of treatment showed (see Supplementary Material, Table 5S) the
m3 of biogas/m3/d), to be compared with the biogas expected strongly granular nature of the sludge which allowed an easier
production as to evaluate the anaerobic degradability of the inlet solid/liquid separation as well as low suspended solids in the
wastewater and, consequently, the achievable energy recovery effluent, in line with similar experimental tests (Lu et al., 2015).
yields. In the first case, the values were compared with the typical Although the excellent performance in terms of organic matter
value of 350 L CH4/kg COD under standard temperature and pres- removal, the UASB effluent was not able to meet the Italian stan-
sure conditions (according to the relation 1, Materials and methods dards for the discharge in superficial water bodies (Table 2), thus
section); in the second case, the biogas production rate was requiring additional treatments herein discussed.
compared with the typical range of 1.5e2.9 as reported in
Fernandes et al. (2010).
3.2. Anoxic-aerobic UF-MBR processes
The observed and theoretical biogas productions were almost
coincident with OLR values lower than 8 kg COD/m3/d, with
The anoxic-aerobic UF-MBR system was started with an initial
measured specific productions in the range of 300e400 L CH4/kg
load of 0.185 kg COD/kgVSS/d, which was gradually decreased ac-
COD and an average value of 320 L CH4/kg COD. The biogas pro-
cording to the trends observed during system operation. The results
duction rate amounted to an average value of 2.6 m3 biogas/m3/d,
in terms of average characteristics of the MBR effluent in corre-
with the highest values observed in correspondence of an OLR of
spondence of each organic loading condition applied as well as in
about 8.5 (although higher values have been reported with OLR
terms of percentage removal, are reported in Table 3 and Fig. 4a,
equal to 11). The average methane content in the biogas was 60%. A
respectively.
further OLR increase, up to 11 kg COD/m3/d, entailed a substantial
The first run (OLR ¼ 0.185 kg COD/kg VSS/d) was characterized
decrease of the observed biogas production (Fig. 2c and d, from day
by COD, TN and N-NHþ 4 removals of 48.0, 67.8 and 81.0%, respec-
15 to day 30). This negative performance was concomitant with the
tively. This run, as discussed below, represented the best opera-
detection, inside the reactor, of persistent whitish foam that caused
tional set up for the biological processes during the entire
significant pressure oscillations as well as frequent overflows (See
investigation. The second run (OLR ¼ 0.100 kg COD/kg VSS/d) was
Fig. 1S, Supplementary material), with a consequent instability of
characterized by COD, TN and N-NHþ 4 removals of 63.1, 30.0 and
the biological processes.
86.4%, respectively. Instead, removals of 58.8, 29.1 and 90.0% for
The optical microscope observations of foam samples (Fig. 3)
COD, TN and N-NHþ 4 , respectively, were observed in the third run
allowed determining the biological nature of the foam, due to the
(OLR ¼ 0.07 kg COD/kg VSS/d). Considering the COD removal, the
presence of large quantities of Yeast-like microorganisms and other
findings were lower than those experienced by Bolzonella et al.
anaerobic filamentous microorganisms that favoured the increase
(2010), Valderrama et al. (2012) and Sheldon and Erdogan (2016),
of the surface tension of the medium and, consequently, the foam
with COD removals of 94, 97 and 90%, respectively. In terms of TN
production. The microscopic investigation showed the presence of
removal, results obtained during the first run were higher than that
significant quantities of Methanosarcina-like microorganisms and
reported in Bolzonella et al. (2010) (54%), mostly related to a higher
Yeast-like microorganism's colonies, inducing to correlate the foam
nitrification efficiency allowed by the modified Ludzack-Ettinger
growth to the application of the maximum OLR (11 kg COD/m3/d)
(MLE) process configuration. The performance achieved in terms
which led to the imbalance between the microbial populations and
of N-NHþ 4 removal during the first run was in line with Sheldon and
the yeast in the reactor. Similar problems were encountered by
Erdogan (2016), who obtained a removal rate of 88.1%. With
Mendes et al. (2006) and Gomes et al. (2011). However, in order to
reference to colour removal, Fig. 4b shows the results by comparing
operate with high OLR as in our case (11 kg COD/m3/d), they
the absorbance detected on (i) MBR influent and effluent samples,
implemented a pre-treatment of the inlet wastewater consisting in
and (ii) on the MBR influent sample previously filtered (indicated as
hydrolysis (Mendes et al., 2006) coupled to enzymatic pre-
MBR filtered influent). A decrease of absorbance in the effluent
treatments (Gomes et al., 2011).
compared to the influent wastewater at all the wavelengths was
Following these evidences, the OLR was preliminarily brought
observed, while the coincidence between the absorbance curves
back at 5e6 kg COD/m3/d and, subsequently, at 4 kg COD/m3/d.
obtained by analysing the inlet filtered wastewater and the MBR
Nevertheless, the only result was the decrease of the foam growing
effluent shows how colour removal was mostly due to the filtration
rate, which was inversely proportional to the applied organic load.
effect of the membrane, with a very limited contribution of the
Therefore, polysiloxaneanti-foam (174 g/l) was directly injected on
biological treatment stage. Most probably, this colouration was due
a weekly basis in the aqueous solution (1:75,000, 5 ml) and in
to the presence of a particular class of compounds, known as
correspondence of the foam, allowing the increase of biogas
melanoidins, originating from the Maillard reaction involving
 L. Petta et al. / Journal of Environmental Management 201 (2017) 177e189 183

Fig. 3. Optic microscope observations of foam and sludge sampled from the UASB reactor and in fresh distillery wastewater, in interdifferential contrast (DIK, 1000 magnification):
(a) Methanosarcina-like microorganisms colonies and presence of Yeast-like microorganisms in foam sample; detail of Methanosarcina-like microorganisms colonies (b) and Yeast-
like microorganisms colonies (c) in foam sample; (d) Methanosarcina-like microorganisms and (e) Yeast-like microorganisms in sludge sample; (f) Yeast-like microorganisms in
fresh influent wastewater.

Table 3
Characteristics of winery distillery wastewater downstream to the anoxic-aerobic UF-MBR process (Average values and standard deviation).

Parameter Unit Organic loading rate (OLR) [kg COD/kg VSS/d] Discharge standardsb

0.185 0.100 0.070

sCOD mg/l 720.0 (±36.0) 510.0 (±35.3) 570.0 (±39.0) 160

Total Nitrogen (TN) mg/l 73.0 (±22.2) 158.9 (±56.0) 161.0 (±52.3) 15
Ammonia Nitrogen (N-NHþ 4) mg/l 21.0 (±7.0) 15.0 (±5.5) 10.0 (±3.8) 15
Nitric Nitrogen (N-NO-3) mg/l n.a.a 11.9 (±4.2) 121.0 (±47.0) 20.0
-
Nitrous Nitrogen (N-NO2) mg/l 3.0 (±1.1) 97.0 (±32.0) n.a.a 0.6
Total Phenols mg/l 55.0 (±6.0) 60.0 (±7.0) 55.0 (±6.0) 0.5
Sulphates (SO2
4 ) mg/l e 200.0 (±21.2) 70.0 (±10.4) 1000
Phosphates (P-PO4) mg/l e 150.0 (±17.2) 154.0 (±18.4) 0
Chlorides (Cl) mg/l e 160.0 (±19.0) 157.0 (±18.2) 1200
Colour e 1:75 1:75 1:75 1:20

Run 1 Run 2 Run 3

a
n.a. ¼ not available.
b
Italian Law (D. Lgs. 152/2006).

residual sugar and amine compounds during the distillation step, as dissolved COD in the aerobic compartment. Fig. 4ceg aim at
found by Nakajima-Kambe et al. (1999), Manisankar et al. (2004) showing the specific trends of the biological processes and the
and Shayegana et al. (2005). Melanoidins compounds are not related operation of the filtration process. In detail, Fig. 4c shows
degradable by traditional biological treatments (Ioannou et al., the trend of COD concentrations in both compartments and in MBR
2015), representing a relevant amount of non-biodegradable COD effluent, highlighting the progressive soluble COD accumulation in
fraction. As above, it justifies the lowest performance, in terms of the aerobic compartment, which values exceeded the total COD
COD removal, compared to that obtained from Sheldon and measured in the influent stream (from day 20). Such occurrence
Erdogan (2016). Considering the other parameters, a percentage was probably related to a retaining action of the UF-membrane in
increase was observed for sulphates (þ200%), chlorides (þ23%) and respect of soluble non-biodegradable COD fractions, which, there-
phosphates (þ4.4%), while total phenols have undergone an fore, were gradually accumulated in the aerobic vessel.
average percentage reduction of 65% (Fig. 4a). Similarly, Fig. 4d shows the trend of nitric (N-NO-3) and nitrous
The progressive OLR reduction from 0.185 to 0.07 kg COD/kg (N-NO-2) nitrogen concentrations in the inlet wastewater as well as
VSS/d was adopted as a consequence of the progressive deteriora- in the aerobic and anoxic compartments. The observed trends show
tion of denitrification and nitrification processes, observed from the that nitrification and denitrification processes initially proceeded
30th day of the experimentation, resulting in a TN removal appropriately during the first run, with a TN removal efficiency
decrease (from 67.8 to 30.0%) and a progressive accumulation of almost equal to 70% (see Fig. 4a); from day 25, a progressive
184 L. Petta et al. / Journal of Environmental Management 201 (2017) 177e189

Fig. 4. Anoxic-aerobic MBR results: (a) Performance in terms of pollutants removal; (b) Spectrophotometric scanning of the MBR feeding and effluent; (c) COD values and (d)
Nitrogen compounds values during the first 60 days of the experimentation; (e) TMP and (f) permeate flux values during the experimentation (120 days); (g) pH values.
 L. Petta et al. / Journal of Environmental Management 201 (2017) 177e189 185

accumulation of nitrites occurred in the aerobic compartment (see

red lines, Fig. 4d) and, later, in the anoxic one (see blue lines,
Fig. 4d), as a result of partial nitrification and denitrification pro-
cesses. The lack of ammonia nitrogen observations in both com-
partments (as explained in Par. 2.4) did not allow further
considerations on specific process kinetics.
Such decrease of performance has probably to be ascribed to a
progressive recalcitrant compounds accumulation, e.g. phenols, in
the aerobic compartment (and, later, in the anoxic one) caused by
the selective retention provided by the UF-membrane, resulting in
the gradual inhibition of nitrification and denitrification processes.
Although no direct phenol detection was provided in the biological
compartments, a moderate retention effect on phenols given by the
UF membrane is likely to be expected, according to Acero et al.
(2005), who observed an apparent rejection coefficient to the
Gallic Acid of more than 20% in experiments carried out with
different UF membranes modules. The consequent progressive in-
crease of phenols concentration in the aerobic vessel may represent
the main cause of the inhibition to nitrification, according to Amor
et al. (2005) who in aerobic batch assays observed an inhibitory
effect to the nitrification process at phenol concentrations ranging
from 100 to 2500 mg/l, leading to the accumulation of nitrite and
ammonium in the reactor.
Due to the low membrane fluxes applied, no severe fouling was
observed during the experiments.
The trend of the transmembrane pressure (TMP) and the specific
permeate flow (JP) during the entire experiment are reported,
respectively, in Fig. 4e and f. Membrane washing operations, rep-
resented with vertical lines, produced a lowering of the TMP
pressure and at the same time an increase of the permeate flow.
During the tests, the average TMP applied (see TMP filtration,
Fig. 4e) was 214.7 mbar while the average value of the JP flow was
0.385 L/m2/h (see Permeate flux, Fig. 4f).
With reference to the third run, the accumulation of inhibitory
substances in the two vessels slightly raised the pH up to values
around 9 (Fig. 4g). The N-NHþ 4 removal observed in the MBR
effluent stream was likely a consequence of the ammonia stripping
favoured by high pH values and the air blown at the bottom of the
tank. The process limitations observed persisted even with more
prolonged periods of biomass adaptation as well as in presence of
lower applied organic load conditions. Following these observa-
tions, the implementation of an additional treatment downstream
to the MBR anoxic-aerobic unit was established, in order to provide
the removal of recalcitrant COD residual fraction as well as colour
and phosphates.

3.3. Chemical precipitation and GAC adsorption as post-treatments

Two different post treatment options, such as chemical precip-

itation and GAC adsorption, were tested as shown in Fig. 1a.
The first tests were carried out in order to evaluate the optimal
lime (Ca(OH)2) dosage. The MBR effluent was treated with
increasing concentrations of lime, variable between 2 and 8 g/l, and
a constant reaction time of 1 min. The maximum efficiency of
phosphates removal (62%), determined immediately after precipi- Fig. 5. Chemical phosphorous removal: (a) Identification of the optimal Ca(OH)2
tation, was observed with a lime dosage of 8 g/l (Fig. 5a). The su- dosage for P-PO4 removal; (b) trend of P-PO4 removal with increasing settling time;
pernatant was collected and analysed also after 5, 7 and 8 days from Trend of (c) total phosphorous removal, (d) P-PO4 removal and (e) COD removal vs.
reaction (contact) time, with a dosage of 8 g Ca(OH)2/l.
the reaction, with Fig. 5b showing the observed increase of phos-
phates removal with higher settling times.
Once the best optimal dosage of 8 g Ca(OH)2/L was identified,
TP removal efficiencies higher than 60%, and a phosphate removal
the effect of the reaction (contact) time (1, 5, 10 and 20 min) on TP,
of about 70% (Fig. 5c and d). Conversely, lime was much less
phosphates and COD removal was investigated, as shown in
effective in COD removal with observed efficiencies always lower
Fig. 5cee. TP and phosphates removal improved as the contact time
than 30% (Fig. 5e). The pH of the treated effluent is on average in the
increased, with values in the range of 64e82% for TP and 64e85%
range 8e9, thus requiring a further correction upstream to the final
for phosphates. The shortest contact time applied (1 min) allowed
186 L. Petta et al. / Journal of Environmental Management 201 (2017) 177e189

discharge. Chemical precipitation was found effective for phosphorus removal

The use of the multi-criteria analysis methodology (De Feo et al., while GAC adsorption resulted as the best option for residual COD
2013) made it possible to identify the optimal operating condition and colour removal. Thus, a combined use of such technologies may
after measuring the process yield in terms of TP, phosphates and represent a suitable solution in order to achieve both the goals of
COD percentage removal. The results of this methodology, starting surface water discharge and agricultural reuse. The use of chemical
from the alternatives matrices of Tables 6S and 7S (see precipitation as a post-treatment on the entire flow to be treated
Supplementary Material) show how 8 g/l of Ca(OH)2 and 15 min entails an overall unit cost of about 5.0 V/m3 for the whole treat-
of reaction time were identified as the optimal condition. ment train (see Table 5), excluding the cost of excess sludge treat-
Further chemical precipitation trials were carried out in order to ment and disposal. Moreover, the anoxic-aerobic UF-MBR
evaluate the colour removal efficiency, seeking to achieve the treatment unit requires a further reduction of the recalcitrant
compliance with discharge standards set by the Italian Law. Tests influent load in order to avoid drawbacks to biological processes.
with lime dosage consisted in evaluating the residual colour after 5, Such considerations suggested the need of further improvements
7 and 10 d contact time with the optimal conditions previously to the treatment chain.
identified. The observed results show that longer contact time Considering the technologies already tested in the present
improved colour removal although the absorbance spectra (Fig. 1S, study, tests aimed at evaluating the benefits given by the adoption
Supplementary Material) showed that the Italian legislation stan- of chemical precipitation with lime upstream and downstream to
dards were only met with 7 d contact time. the UASB reactor were performed. In such tests, a lime dosage of
All the experiments using activated carbon resulted in a non- 8 g/l was hired as detailed in Par. 3.3.
detectable colour with 1:20 dilution at the effluent, confirming a As shown in Table 6, lime dosage allowed high phosphorous
high effectiveness in colour removal. The activated carbon treat- compounds (97e100%) and COD removals (67e68%) in raw
ment unit also resulted in a 95% COD removal, whereas no signif- wastewater, thus improving the performance observed by Braz
icant TP and phosphate removal was observed with values of 3% et al. (2010) (COD removal ¼ 18.9e29.7%). Furthermore, lime
and 4%, respectively. The performances of the two treatment op- dosage allowed the removal of part of the recalcitrant COD in the
tions are reported in Table 4. UASB effluent, with an expected improvement of the downstream
The comparison highlighted the suitability of chemical precip- biological processes. Consequently, the combined adoption of
itation to remove phosphorous compounds while the colour was anoxic-aerobic UF-MBR and activated carbons (resulted as the most
removed only after high sedimentation days (7 days). Instead, the effective method for colour removal) can be considered as an
adsorption on GAC was suitable for recalcitrant COD and colour appropriate solution for the compliance with the effluent legal
removal. standards. Based on these considerations, the most appropriate
distillery wastewater treatment cycle should include the following
3.4. On the optimization of the investigated treatment chain unit processes: (i) Chemical precipitation with lime; (ii) UASB
treatment; (iii) Chemical precipitation of UASB effluent; (iv)
Table 5 shows the material balances referred to the main pa- Anoxic-aerobic UF-MBR; (v) GAC adsorption. The corresponding
rameters such as COD, TN and TP/phosphates. Additionally, the unit costs of the entire treatment train is estimated to be about 6.3
table summarizes the optimal operating conditions of biological V/m3, which could be decreased up to 5.0 V/m3 if the lime dosage is
processes (in terms of OLR) as well as the main dysfunctions provided only to a partial amount of the total flow-rate upstream
identified during the trial. and downstream to UASB process unit. However, such option re-
The inlet wastewater, characterized by acidic pH, required a quires further experimental investigations aimed at verifying the
preliminary correction in order to obtain a value of 6.2. The feeding specific removal performances in order to guarantee the compli-
of UASB process was optimal for an OLR value of 4.5 kg COD/m3/d. ance with the final effluent standards.
The adoption of higher values resulted in a higher risk of biological Among the “external technologies”, not tested in the present
foams formation with consequent negative effects on biogas pro- experimentation, struvite precipitation in UASB effluent can indeed
duction and process stability. be considered a suitable option to achieve nitrogen recovery, also in
The anoxic-aerobic UF-MBR system allowed the removal most of order to unload the biological units expected downstream. In
the organic load and almost all of the nitrogen compounds. The particular, anaerobically pre-treated wastewater is the most suit-
feeding was optimal with an OLR of about 0.185 kgCOD/kg VSS/d. able for struvite crystallization, due to the previous ammonification
Although the UF-membrane determined a better solid/liquid sep- of the organic nitrogen with ammonium formation (Altinbaş,
aration with an almost complete suspended solids removal, it 2009). Temperature, pH, TSS and wastewater chemical composi-
contributed to a progressive accumulation in the basins of recal- tion are among the main parameters to be assessed for an optimal
citrant compounds that negatively affected the yields of biological struvite precipitation process. Generally, optimum pH values are in
processes (as resulted during the long-term MBR treatment tests). the range 8e11. However, further site-specific experimentations
The presence of residual COD as well as other compounds such need to be carried out at laboratory scale.
as phosphates, chlorides, sulphates and the colour in the MBR Table 5 shows further aspects to be optimized for a successful
effluent made it necessary an additional finishing treatment. technological scale-up: a) the specific flow of permeate to be

Table 4
Performance comparison between chemical precipitation with lime and GAC adsorption as post-treatments for colour removal.

Processes Performance [Percentage removal]

%COD %TP %P-PO4 Colour removalb

Chemical precipitation with limea 31.2 79.6 85.3 Complete after 7-days
GAC adsorption 95.0 3.0 4.0 Complete
a
Optimal conditions: dosage of 8 g/L of Ca(OH)2 and 15-min of reaction time.
b
Complete ¼ Not perceptible with 1:20 dilution.
Table 5
Material balances referred to the main contaminants and to the entire treatment chain, costs and main remarks.

N. Description Main process parameters Volumetric Mass flow rate of pollutants [g/d] Percentage removal [%] Biogas Sludge Remarks Unit
Section flow rate [l/ production production costs
a COD TN N-NHþ
4 COD TN N-NHþ
4
d] [LCH4/kgCOD] [cm3/d] [V/m3]b

1 UASB feeding 1 e 1.61 72.0 0.410 0.218 e e e e e Acidic pH requiring e

UASB feeding 2 e 2.42 108.0 0.615 0.327 e e e e e a correction with
UASB feeding 3 e 6.19 276.0 1.572 0.835 e e e e e soda.

L. Petta et al. / Journal of Environmental Management 201 (2017) 177e189

2 UASB effluent 1 OLR1 ¼ 3 kgCOD/m3/d 1.61 1.766 0.355 0.034 97.5 13.4 84.0 350 e Biological foams 0.10
UASB effluent 2 OLR2 ¼ 4.5 kgCOD/m3/d 2.42 3.051 0.581 0.363 97.2 5.5 11.1 400 e with OLR >6.2
UASB effluent 3 OLR3 ¼ 11.5 kgCOD/m3/d 6.19 11.077 1.368 0.990 96.0 13.0 18.5 100 e 3
kgCOD/m /d;
Presence of colour.
3 MBR feeding 1 e 1.03 1.428 0.234 0.114 e e e e e e e
MBR feeding 2 e 0.55 0.764 0.125 0.061 e e e e e
MBR feeding 3 e 0.38 0.531 0.087 0.042 e e e e e
4 MBR effluent 1 OLR1 ¼ 0.185 kgCOD/kgVSS/d 1.03 0.743 0.075 0.022 48.0 67.8 81.0 e 70c Accumulation of 0.28
MBR effluent 2 OLR2 ¼ 0.100 kgCOD/kgVSS/d 0.55 0.282 0.088 0.008 63.1 30.0 86.4 e 33c recalcitrant
MBR effluent 3 OLR3 ¼ 0.070 kgCOD/kgVSS/d 0.38 0.219 0.062 0.004 58.8 29.1 90.9 e 15c compounds.
Presence of colour.
Need to reduce the
organic load and
the ammonia
toxicity.
Low JP flow.
Post-treatments COD TP P-PO4 COD TP P-PO4
5 Feeding e 5.00 5.130 1.275 0.956 e e e e e
6 Adsorption effluent e 1.00 3.530 0.261 0.141 95.0 3.0 4.0 e e High colour and 3.5
COD removals.
6 bis Precipitation Best condition: (i) Ca(OH)2 5.00 3.530 0.261 0.141 31.2 79.6 85.3 e 900e1400 High sludge 1.2
effluent dosage ¼ 8 g/l; Contact production;
time ¼ 15 min. Limited colour
removal.
High contact time.
Need to reduce the
pH.
a
See also Fig. 1.
b
Operating costs without sludge treatment and disposal. Assumptions for the adsorption treatment: Consumption of GAC ¼ 3.5 kg/m3; Cost of GAC ¼ 1.0 V/kg dry-product; Assumptions for the chemical precipitation: Cost of
lime ¼ 0.15 V/kg; Lime dosage ¼ 8 kg/m3. In terms of energy consumption, the main item is due to the MBR system with a values in the range 2.0e3.6 kWh/m3. Considering the MBR, the main items are aeration (58%) and
membranes (~32%).
c
Average values.

187
188 L. Petta et al. / Journal of Environmental Management 201 (2017) 177e189

Table 6 Dr. Carmela Cellamare and Dr. Loredana Stante for providing the
Results related to the chemical precipitation with lime upstream and downstream optic microscope observations of foam and sludge samples from the
the UASB reactor (“internal optimization” of the treatment chain).
UASB reactor. The Authors equally acknowledge the four anonymous
Sample Dosage of Ca(OH)2 [g/L] Reaction time [min] Performance reviewers for their useful suggestions that helped to improve the
[Percentage paper.
removal]

%P-PO4 %COD
Appendix A. Supplementary data
Test 1: Chemical precipitation upstream the UASB
1 8 5 97% 67% Supplementary data related to this article can be found at http://
2 8 10 98% 68%
3 8 20 100% 67%
dx.doi.org/10.1016/j.jenvman.2017.06.042.
Test 2: Chemical precipitation downstream the UASB
4 8 5 80% 46% References
5 8 10 90% 49%
6 8 20 65% 50% Acero, J.L., Benitez, J.F., Leal, A.I., Real, F.J., 2005. Removal of phenolic compounds in
water by ultrafiltration membrane treatments. J. Environ. Sci. Heal. A 40,
1585e1603.
Akunna, J.C., Clark, M., 2000. Performance of a granular-bed anaerobic baffled
applied, since the adopted value (0.385 L/m2/h) during the exper- reactor (GRABBR) treating whisky distillery wastewater. Biores. Technol. 74,
imental tests would require a high membrane area or a large 257e261.
Altinbaş, M., 2009. Nitrogen recovery via struvite production. In: Cervantes, F.J.
storage tank; b) the HRT in the contact tank for chemical precipi- (Ed.), Environmental Technologies to Treat Nitrogen Pollution. IWA Publishing,
tation with lime, since the optimal value achieved (7 days) would London, pp. 239e268.
require excessive contact volumes. Therefore, future investigations Amaral-Silva, N., Martins, R.C., Paiva, C., Castro-Silva, S., Quinta-Ferreira, R.M., 2016.
A new winery wastewater treatment approach during vintage periods inte-
are expected as to take into account the need of full-scale treatment
grating ferric coagulation, Fenton reaction and activated sludge. J. Environ.
as argued above. Chem. Eng. 4 (2), 2207e2215.
Amor, L., Eiroa, M., Kennes, C., Veiga, M.C., 2005. Phenol biodegradation and its
effect on the nitrification process. Water Res. 39, 2915e2920.
4. Conclusions
Andreottola, G., Foladori, P., Ziglio, G., 2009. Biological treatment of winery
wastewater: an overview. Water Sci. Technol. 60 (5), 1117e1125.
The investigated treatment scheme considered the UASB treat- Andreottola, G., Foladori, P., Nardelli, P., Denicolo, A., 2005. Treatment of winery
ment unit as a pre-treatment aimed at reducing the easily biode- wastewater in a full-scale fixed bed biofilm reactor. Water Sci. Technol. 51 (1),
71e79.
gradable COD fraction and leading to methane production. The Andreottola, G., Foladori, P., Ragazzi, M., Villa, R., 2002. Treatment of winery
optimal operational set up was reached at OLR of 4.5 kgCOD/m3/ wastewater in a sequencing batch biofilm reactor. Water Sci. Technol. 45 (12),
d with a correspondent COD and total phenols removal efficiency of 347e354.
Andreottola, G., Nardelli, P., Nardin, F., 1998. Demonstration plant experience of
97.2 and 75.0%, respectively. Higher OLR values (in the range of winery wastewater anaerobic treatment in a hybrid reactor. In: Second Inter-
6.2e11.5 kg COD/m3/d) resulted in biological yeasts foams forma- national Specialized Conference on Winery Wastewater. May 5e7. Cema-
tion with a consequent decrease of biogas production and limited grefeDICOVA, Bordeaux (France).

Basset, N., Santos, E., Dosta, J., Mata-Alvarez, J., 2016. Start-up and operation of an
process stability. AnMBR for winery wastewater treatment. Ecol. Eng. 86, 279e289.
The anoxic-aerobic UF-MBR unit was aimed at providing re- Beck, C., Prades, G., Sadowski, A.-G., 2005. Activated sludge wastewater treatment
sidual COD, nitrogen and colour removal. The optimal operational plants optimisation to face pollution overloads during grape harvest periods.
Water Sci. Technol. 51 (1), 81e88.
set up was reached at OLR of 0.185 kg COD/kg VSS/d, with a Bolzonella, D., Fatone, F., Pavan, P., Cecchi, F., 2010. Application of a membrane
correspondent COD, TN, ammonia nitrogen and total phenols re- bioreactor for winery wastewater treatment. Water Sci. Technol. 62 (12),
movals of 48.0, 67.8, 81.0 and 65.0%, respectively. The adoption of 2754e2759.
Braz, R., Pirra, A., Lucas, M.S., Peres, J.A., 2010. Combination of long term aerated
UF-membranes as well as the presence, in the UASB effluent, of
storage and chemical coagulation/flocculation to winery wastewater treatment.
recalcitrant compounds (i.e., phenols), led to their accumulation in Desalination 263, 226e232.
the aerobic and anoxic compartments, resulting in the inhibition of Brucculeri, M., Bolzonella, D., Battistoni, P., Cecchi, F., 2005. Treatment of mixed
biological processes and relying the treatment efficiency on the municipal and winery wastewaters in a conventional activated sludge process:
a case study. Water Sci. Technol. 51 (1), 89e98.
membrane filtration effect only. Buntner, D., Sanchez, A., Garrido, J.M., 2013. Feasibility of combined UASB and MBR
Chemical precipitation with lime as post-treatment allowed TP, system in dairy wastewater treatment at ambient temperatures. Chem. Eng. J.
phosphates and COD removals of 79.6, 85.3 and 31.2%, respectively 230, 475e481.
Chai, S., Guo, J., Chai, Y., Cai, J., Gao, L., 2014. Anaerobic treatment of winery
(8 g/l lime dosage and 15 min reaction time) while the GAC wastewater in moving bed biofilm reactors. Desalin. Water Treat. 52 (10e12),
adsorption confirmed a high effectiveness in colour removal. 1841e1849.
The criticalities observed during the experimentation high- Council Directive 91/271/EEC of 21 May 1991 Concerning Urban Waste-water
Treatment.
lighted the important role of chemical processes for the optimiza- De Gisi, S., Galasso, M., De Feo, G., 2013. Full-scale treatment of wastewater from a
tion of the whole treatment chain: the use of a chemical biodiesel fuel production plant with alkali-catalyzed transesterification. Envi-
precipitation step on the UASB effluent appears a suitable solution ron. Technol. 34 (7), 861e870.
De Feo, G., Galasso, M., Landi, R., Donnarumma, A., De Gisi, S., 2013. A comparison of
to reduce the recalcitrant COD content, thus facilitating the long- the efficacy of organic and mixed-organic polymers with polyaluminium
term sustainability of the anoxic-aerobic UF-MBR treatment, with chloride in chemically assisted primary sedimentation (CAPS). Environ. Technol.
a corresponding overall unit costs between 5.0 and 6.3 V/m3 34 (10), 1297e1305.
De Gisi, S., Petta, L., Farina, R., De Feo, G., 2014. Using a new incentive mechanism to
depending on the amount of the flowrate upstream and down-
improve wastewater sector performance: the case study of Italy. J. Environ.
stream to UASB process unit to be treated with lime dosage. Manag. 132, 94e106.
De Gisi, S., Lofrano, G., Grassi, M., Notarnicola, M., 2016. Characteristics and
adsorption capacities of low-cost sorbents for wastewater treatment: a review.
Acknowledgements
Sustain. Mat. Technol. 9, 10e40.
D. Lgs 152/2006 (Italian Legislative Decree No. 152/2006). Decreto Legislativo 3
The study was carried out within the TECNOPOLE Programme Aprile 2006, n. 152. “Norme in materia ambientale”, Gazzetta Ufficiale n. 88 del
funded by the Emilia-Romagna Region, Italy. The Authors acknowl- 14 Aprile 2006 e Supplemento Ordinario n. 96 (in Italian).
Farina, R., Cellamare, C.M., Stante, L., Giordano, A., 2004. Pilot scale anaerobic
edge Dr. Martina Puccetti, Dr. Daniela Saturno and Dr. Alessandro sequencing batch reactor for distillery wastewater treatment. In: Proceedings of
Spagni for the valuable support during all the experimental trials and the 10th World Congress Anaerobic Digestion 2004: Anaerobic Conversion for
 L. Petta et al. / Journal of Environmental Management 201 (2017) 177e189 189

Sustainability, from 29 August to 2 September, Montreal, Canada. wastewaters. Water Sci. Technol. 56, 89e94.
Fernandez, B., Seijo, I., Ruiz-Filippi, G., Roca, E., Tarenzi, L., Lema, J.M., 2007. Char- Mulidzi, A.R., 2010. Winery and distillery wastewater treatment by constructed
acterization, management and treatment of wastewater from white wine pro- wetland with shorter retention time. Water Sci. Technol. 61, 2611e2615.
duction. Water Sci. Technol. 56 (2), 121e128. Mulidzi, A.R., 2007. Winery wastewater treatment by constructed wetlands and the
Fernandes, B., Peixoto, G., Rui Albrecht, F., Saavedra del Aguila, N.K., Zaiat, M., 2010. use of treated wastewater for cash crop production. Water Sci. Technol. 56 (2),
Potential to produce biohydrogen from various wastewaters. Energy Sustain. 103e109.
Dev. 14, 143e148. Nakajima-Kambe, T., Shimomura, M., Nomura, N., Chanpornpong, T., Nakahara, T.,
Gomes, D.R.S., Papa, L.G., Cichello, G.C.V., Belançon, D., Pozzi, E.G., Balieiro, J.C.C., 1999. Decolourization of molasses wastewater by Bacillus sp. under thermo-
Monterrey-Quintero, E.S., Tommaso, G., 2011. Effect of enzymatic pretreatment philic and anaerobic conditions. J. Biosci. Bioeng. 87 (1), 119e121.
and increasing the organic loading rate of lipid-rich wastewater treated in a Pathe, P.P., Rao, N.N., Kharwade, M.R., Lakhe, S.B., Kaul, S.N., 2002. Performance
hybrid UASB reactor. Desalination 279, 96e103. evaluation of full scale effluent treatment plant for distillery spent wash. Intern.
Hussain, A., Kumar, P., Mehrotra, I., 2008. Treatment of phenolic wastewater in J. Environ. Stud. 59 (4), 415e437.
UASB reactor: effect of nitrogen and phosphorous. Biores. Technol. 99, Petropoulos, E., Cuff, G., Huete, E., Garcia, G., Wade, M., Spera, D., Aloisio, L.,
8497e8503. Rochard, J., Torres, A., Weichgrebe, D., 2016. Investigating the feasibility and the
Ioannou, L.A., Li Puma, G., Fatta-Kassinos, D., 2015. Treatment of winery wastewater limits of high rate anaerobic winery wastewater treatment using a hybrid-EGSB
by physicochemical, biological and advanced processes: a review. J. Hazard. bio-reactor. Process Saf. Environ. 102, 107e118.
Mater 286, 343e368. Petruccioli, M., Duarte, J.C., Eusebio, A., Federici, F., 2002. Aerobic treatment of
Lu, X., Zhen, G., Ledezma Estrada, A., Chen, M., Ni, J., Hojo, T., Kubota, K., Li, Y.-Y., winery wastewater using a jet-loop activated sludge reactor. Process Biochem.
2015. Operation performance and granule characterization of upflow anaerobic 37, 821e829.
sludge blanket (UASB) reactor treating wastewater with starch as the sole Rodríguez-Chueca, J., Amor, C., Silva, T., Dionysiou, D.D., Li Puma, G., Lucas, M.S.,
carbon source. Biores. Technol. 180, 264e273. Peres, J.A., 2017. Treatment of winery wastewater by sulphate radicals: HSO5/
Lucas, M.S., Peres, J.A., Puma, G.L., 2010. Treatment of winery wastewater by ozone- transition metal/UV-A LEDs. Chem. Eng. J. 310 (2), 473e483.
based advanced oxidation processes (O3, O3/UV and O3/UV/H2O2) in a pilot- Serrano, L., de la Varga, D., Ruiz, I., Soto, M., 2011. Winery wastewater treatment in a
scale bubble column reactor and process economics. Sep. Purif. Technol. 72, hybrid constructed wetland. Ecol. Eng. 37, 744e753.
235e241. Shayegana, J., Pazoukib, M., Afshari, A., 2005. Continuous decolourization of
Manisankar, P., Rani, C., Viswanathan, S., 2004. Effect of halides in the electro- anaerobically digested distillery wastewater. Process Biochem. 40, 1323e1329.
chemical treatment of distillery effluent. Chemosphere 57, 961e966. Sheldon, M.S., Erdogan, I.G., 2016. Multi-stage EGSB/MBR treatment of soft drink
Mendes, A.A., Pereira, E.B., Castro, H.F., 2006. Effect of the enzymatic hydrolysis industry wastewater. Chem. Eng. J. 285, 368e377.
pretreatment of lipids-rich wastewater on the anaerobic biodigestion. Biochem. Standard Methods for the Examination of Water and Wastewater, nineteenth ed.,
Eng. J. 32, 185e190. 1995. American Public Health Association/American Water Works Association/
Metcalf & Eddy, 2003. Wastewater Engineering. Treatment and Reuse, fourth ed. Water Environment Federation, Washington, DC.
McGraw Hill, New York. Thanikal, J.V., Torrijos, M., Habouzit, F., Moletta, R., 2007. Treatment of distillery
Moletta, R., 2005. Winery and distillery wastewater treatment by anaerobic vinasse in a high rate anaerobic reactor using low density polyethylene sup-
digestion. Water Sci. Technol. 51 (1), 137e144. ports. Water Sci. Technol. 56 (2), 17e24.
Molina, F., Ruiz-Filippi, G., García, C., Roca, E., Lema, J.M., 2007. Winery effluent Valderrama, C., Ribera, G., Bahí, N., Rovira, M., Gime nez, T., Nomen, R., Lluch, S.,
treatment at an anaerobic hybrid USBF pilot plant under normal and abnormal Yuste, M., Martinez-Llado , X., 2012. Winery wastewater treatment for water
operation. Water Sci. Technol. 56 (2), 25e31. reuse purpose: conventional activated sludge versus membrane bioreactor
Mosteo, R., Ormad, M.P., Ovelleiro, J.L., 2007. Photo-Fenton processes assisted by (MBR). A comparative case study. Desalination 306, 1e7.
solar light used as preliminary step to biological treatment applied to winery