Efficient solvent-less separation of lipids from municipal wet sewage scum and their sustainable conversion into biodiesel

Efficient solvent-less separation of lipids from municipal wet sewage scum and their sustainable conversion into biodiesel

Renewable Energy 90 (2016) 55e61 Contents lists available at ScienceDirect Renewable Energy journal homepage: www.elsevier.com/locate/renene Efficie...

530KB Sizes 3 Downloads 63 Views

Renewable Energy 90 (2016) 55e61

Contents lists available at ScienceDirect

Renewable Energy journal homepage: www.elsevier.com/locate/renene

Efficient solvent-less separation of lipids from municipal wet sewage scum and their sustainable conversion into biodiesel Luigi di Bitonto a, Antonio Lopez b, Giuseppe Mascolo a, Giuseppe Mininni b, Carlo Pastore a, * a b

CNR-IRSA, via De Blasio 5, 70132, Bari, Italy CNR-IRSA, via Salaria km 29,300, 00015, Montelibretti, Roma, Italy

a r t i c l e i n f o

a b s t r a c t

Article history: Received 15 September 2015 Received in revised form 26 November 2015 Accepted 19 December 2015 Available online xxx

A very efficient separation of lipids from wet sewage scum was optimised and positively tested on samples taken from several wastewater treatment plants (WWTPs). By simply heating sewage scum at 353 K and centrifuging the heated mass at 4000 rpm per 1 min, a recoverability of 93e99% of total oils was always obtained. This procedure resulted to be effective on samples with very different starting water contents. In all cases, extracted lipids have a very similar composition in terms of free fatty acids (FFAs), calcium soaps (32e40%wt) and glycerides (mono-, di- and tri-glycerides were practically absents), as well as fatty acid profiles. Once separated, lipids were converted into biodiesel through a direct esterification process carried out by adopting three sequential batch reactors, in which methanol and catalysts were charged in counter current. In this way, the complete conversion (>99%) of starting FFAs into FAMEs was perfectly matched with using the minimum amount of reactants under very mild conditions (345 K, 2 h). The overall convenience of the process was completed by the anaerobic digestion of fibrous residues obtained from centrifugation of starting sewage scum: the final biogas resulted largely enough to sustain the heat of process. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Sewage scum Biodiesel Centrifugation Direct esterification FAMEs

1. Introduction Production of Fatty Acid Methyl Esters (FAMEs), namely biodiesel, will remain one of the main important European targets in terms of renewables in transport fuels for future [1]. In fact, the value of 10% of renewable fixed by European Parliament still remains the final objective to be achieved by 2020. In addition, new forthcoming rules have been introducing which strongly consider also the sustainability of the renewable considered. More specifically, in order to limit the use of renewables with a high environmental impact and to better regulate the indirect land use change due to the cultivation of crop-for-fuel, a cap for “first generation” biodiesel is going to be fixed to 7%, ever more privileging the use of alternative sources such as waste cooking vegetal oils, animal fats, and alternatives cleaner energy sources (renewable electricity for rails and cars) [2]. In this context, sewage scum could represent a really innovative

* Corresponding author. E-mail address: [email protected] (C. Pastore). http://dx.doi.org/10.1016/j.renene.2015.12.049 0960-1481/© 2015 Elsevier Ltd. All rights reserved.

source which fully respect these criteria of sustainability. Simply recoverable from wastewater treatment plants (WWTPs) through a physical separation (by using a preliminary floatation) operated on urban sewages, however it is not always configured in WWTPs. Nowadays, separation of sewage scum is often avoided to prevent the respective successive disposal, which should represent a further cost of 200e240 Euro ton1. As a matter of fact, sewage scum are in some cases separated in preliminary treatments together with raw solids (sands, stones, etc.) or more frequently directly sent to subsequent sewage treatments, where physical decantation and/or biological oxidation occur without plenty exploit its real potential. Mentioned to be co-fed in anaerobic digestion [3], as already carried out on fat oils and grease recovered from trap-grease [4,5], the valorisation of the lipid component of such a kind of waste to obtain biodiesel has been ever more investigated [6]. The potential use of sewage scum as a source of biodiesel was recently confirmed by using a two-step process of extraction, with hexane and acetone on dry samples and successive conversion of lipids into biodiesel [7]. Very interestingly, based on the same two-step scheme, an alternative solvent-less process of separation of lipids from sewage

56

L. di Bitonto et al. / Renewable Energy 90 (2016) 55e61

scum was studied, by simply adding a mineral acid into wet sewage scum [8]. When the overall mixture was heated to 333 K, a spontaneous efficient separation of an oily phase was achieved and its successive conversion into biofuel was carried out. Finally, even in situ reaction of wet sewage scum with sulphuric acid and methanol (with the eventual co-presence of hexane) was found to efficiently convert starting Free Fatty Acids (FFAs) into FAMEs [9]. All these approaches suffer from the limitations of contaminating the starting sewage scum. So that, besides formation of biodiesel, at the end of the process, a new waste is always generated which consists into residual sewage scum wet of organic solvent, or mineral acids or both. Management of this new special waste could damper the economic balance of the process and in any case, the biodiesel obtained from these processes could have a too high environmental impact. In this work, a simple centrifugation of the heated sewage scum was proposed to recover lipid fraction without adding any solvents and neither acids. After optimisation study of speed, temperature and time of centrifugation, the best separating conditions were then adopted for samples up-taken from different WWTPs. A detailed analysis of recoverability and qualitative nature of the extracts were conducted in order to define the applicability of this process to different sewage scum. The effect of the starting water content was also evaluated. Then, after chemical activation of the separated lipids operated with formic acid, the oily phase was reacted with AlCl3$6H2O methanol solution, by adopting optimised reaction conditions of direct esterification [10]. In detail, for the first time, a new configuration of three sequential batch reactors in counter current was tested for converting lipids of sewage scum into FAMEs, minimising the amount of reactants and maximising the FFAs conversion over the thermodynamic limit. 2. Materials and methods 2.1. Reagents and instruments Hexane and methyl-heptadecanoate were SigmaeAldrich pure grade reagents (99%). Methanol and AlCl3$6H2O were purchased by Baker, whereas formic acid, potassium hydroxide, diethyl-ether, ethanol, nitric acid and hydrochloric acid were Carlo Erba pure reagents. A Rotofix 32 Hettich Centrifuge was used for centrifugation experiments. Gas chromatographic determinations were performed using a Varian 3800 GC-FID, whereas a Perkin Elmer Clarus 500 gaschromatograph interfaced with a Clarus 500 spectrometer was used (GCeMS) for qualitative identification of different species. Both instruments were configured for cold on-column injections, using a HP-5MS capillary column (30 m; 0.32 mm, 0.25 mm film). Agilent 7700 Inductively Coupled Plasma-Mass spectrometer equipped with a collision cell used in He mode was used for metals analysis. Perkin Elmer Spectrum BX was used for recording FTIR spectrum by using rectangular Perkin Elmer KBr cells for liquid samples. Sugars were determined by using a GS50 chromatography system (Dionex-Thermo Fisher Scientific, Sunnyvale, CA, USA) equipped with an AS50 autosampler and an ED50 pulsed amperometric detector (gold electrode). Microfiltered water solutions were injected via a 25 mL loop, into a Carbopac PA10 Analytical column (250 mm, 4 mm; Dionex) using a flow of 1.0 mL min1 of an aqueous gradient of KOH generated by an EG40 Eluent generator, equipped with an EGC KOH III cartridge (KOH concentration profile:18 mM for 30 min, 200 mM for 10 min and again 18 mM for 15 min). Glucose was used as standard for the determination and quantification of cellulose.

2.2. Sewage scum Samples of sewage scum were collected from WWTPs of Bari West (242000 Population Equivalent, PE), Andria (100000 PE), Barletta (90000 PE), Bisceglie (50000 PE), Putignano (12000 PE) and Polignano a Mare (12000 PE), all located in South of Italy. All these different dimensioned WWTPs, mainly process urban sewages. Samples were directly analysed for determining total solids, ashes, total lipids (which was further characterised in terms of FFAs, soaps and fatty acids profile), cellulose, lignin and proteins [11]. Then, they were immediately processed (within two days) avoiding long time of conservation (277 K). 2.2.1. Determination of total solids and ashes Determination of total solids (TS) was performed according to the ISO 11465 method [12]. Samples of sewage scum (20 g) were placed in an oven at 378 K for 24 h until to obtain a constant weight. The results were expressed as weight percentage (%wt) of residual solid after thermal treatment with respect to the starting material. Then, the dried samples were heated in a muffle furnace at 823 K for 2 h and the respective ashes were also determined. 2.2.2. Determination of total lipids Wet sewage scum (20 g) was placed in a Falcon tube of 50 mL with 20 mL hexane and shaken for 2 min at room temperature. A bi-phasic system was obtained: (i) an upper organic suspension and a (ii) lower phase of wet sewage scum. Then, organic suspensions were separated and recovered. This procedure was repeated four times and all organic fractions were collected into a unique flask. The solvent was then evaporated under nitrogen flow and the final residue was gravimetrically quantified and characterized (FFAs, Soaps, FFAs profile, Average Molecular Weight (AMW), glycerides, sterols and waxes [13]). 2.2.3. FFAs and soaps determination FFAs were determined via titration using a 0.1 N KOH normalised solution and phenolphthalein as indicator in a 1:1 diethylether:ethanol medium (1 g sample dissolved into 150 mL solvent). Soaps were determined by titration using a 0.1 M HCl normalised solution and methyl-red as indicator in a 1:1 diethylether:ethanol medium (1 g sample dissolved into 150 mL solvent). Weight percentages were estimated using the AMW determined gas-chromatographically. Soaps content was also quantified through a gravimetric determination: raw oily sample (3 g) was suspended into hexane (20 mL). The white-grey solid which was separated by centrifugation, was then washed with further hexane (3  5 mL) and dried under nitrogen flow before to be gravimetrically determined. 2.2.4. Fatty-acid profile and AMW determination Fatty-acid profile was determined by weighing 20 mg of raw sample into a glass reactor of 5 mL, together with 2 mL toluene, methanol and concentrated H2SO4 (2:2:0.01 v:v:v). The tube was closed and then placed into an ultrasonic bath and kept at 343 K for 5 h. Then, 1 mL of 1000 ppm methyl-heptadecanoate toluene solution was added (as internal standard) and 1 mL was injected to the gas-chromatograph. AMW was calculated using the GC-FID analysis data and applying the following equation:

P AMW ¼

Ai MWi P Ai

(1)

where Ai and MWi are the area (gas-chromatographically detected) and the molecular weight of an identified FFA, respectively.

L. di Bitonto et al. / Renewable Energy 90 (2016) 55e61

2.2.5. Cellulose and lignin NREL method for “Determination of Structural Carbohydrates and Lignin in Biomass” was partially adapted [14] and applied on samples from which lipids were already separated. Dry solids (100 mg) were dissolved and kept under agitation at 303 K for 60 min in 3 mL of 72% sulphuric acid. Then, with 84 mL of distilled water, the acid solution was transferred into a 250 mL glass balloon and kept under reflux for 1 h. The suspension was cooled and filtered on a filtering crucibles previously prepared and weighted. The filtered solution was diluted 50 times (20 mL into 1 mL of mQ water) and analysed for sugars determinations. On the other side, filtered solids were abundantly washed with water and dried for 24 h at 378 K and then weighted. Insoluble lignin was calculated by the difference between this weight and the respective ashes obtained after putting the same filtering crucible into an oven at 823 K for 3 h.

57

carbonyl band of FFA at 1710 cm1 was revealed, whereas neither of carbonyl bands of calcium soaps (1576 and 1538 cm1) and calcium formate (1600 cm1) were detected [15]. 2.5. Reactive conditions

Sewage scum (30 g) was placed and closed in a Falcon tube of 50 mL and heated in a thermostatic water bath (343 or 353 K). After this thermal treatment, samples were rapidly centrifuged (at 4000 or 3000 rpm) for a defined time (1 or 2 min) to obtain a threephasic system (Fig. 1): (i) an upper brown oily phase (liquid at 353 K, but solid at room temperature) constituted by the lipid fraction, (ii) an aqueous intermediate phase and (iii) a lower phase of wet solid residues. Upper phase and residual decanted solids were recovered and dried at 378 K for 24 h. Lipids were characterized in order to define FFAs, soaps and FAs profile, whereas the solids were dried, quantified and analysed for determining ashes, cellulose, lignin and proteins contents.

Activated oil (20 g) was charged into a 100 mL Pyrex reactor with 20 g methanol solution of AlCl3$6H2O (3 g in 200 g of solvent) with the aim of obtaining a final molar percentage of aluminium chloride of 2% respect to starting FFAs. The reactor was placed into a thermostatic bath at 345 K and was kept on agitation by using a magnetic stirrer for 2 h. Reacting mixture was cooled and allowed to be separated into a light methanol layer (M1) and an oily phase (O1). Then, M1 methanol phase was used for feeding a new reactor charged with fresh activated oil (obtaining at the end M2 and O1), whereas O1 oily phase was reacted with fresh aluminium chlorides methanol solution in a different reactor (determining the formation of O2 and M1). After this cycle of reaction, three reactors were fed (Fig. 2): R1, in which methanol already used two times was again reacted with 20 g fresh activated oil, R2 in which O1 was reacted with M1, and R3 in which O2 was reacted with fresh methanol solution. In this way, an overall system of 3 sequential batch reactors was built up, in which after every 2 h, 20 g of fresh activated oil and 20 g AlCl3$6H2O methanol solution were fed respectively in R1 and R3, and exhausted methanol and final oil products (M3 and O3, reacted both for three times) were separated always from R1 and R3. 14 cycles of reaction were conducted. Final exhausted methanol (M3) and oily products (O3) were analysed in terms of methanol (loss of weight through nitrogen flow), water (loss of weight at 378 K for 24 h), aluminium, FFAs and FAMEs content, in order to fully characterise the overall system.

2.4. Activation of lipid phase

2.6. Aluminium analysis

Lipids recovered through centrifugation (20 g) were reacted with the stoichiometric amount of (3e4 g) formic acid calculated respect to the starting soaps. The overall mixture was kept under agitation at 333 K and immediately centrifuged at 4000 rpm per 2 min. Activated lipids (16e17 g, 80e85%) were then separated from the decanted solids and used for their successive conversion into FAMEs. Completeness of the reaction and efficacy of separation were confirmed through FTIR analysis of the activated oil: only

About 5 g sample were mineralised with 20 mL of 65% HNO3 and heated up to formation of red-brown fumes. Then, further HNO3 (5 mL) was added and heated again. These operations were repeated (at least 5 times) until no oxidation was apparent. Final acid residue was then reduced to a volume of 2e3 mL and brought to an exact volume using distilled water. Aqueous samples were analysed using an Agilent 7700 Inductively Coupled Plasma-Mass spectrometer.

2.3. Centrifugation of heated sewage scum

3. Results and discussion 3.1. Sewage scum characterisation and lipid separation and activation

Fig. 1. Separation of phases on raw sewage scum: 1) starting sewage scum at room temperature, 2) sewage scum heated at 353 K and 3) sewage scum heated at 353 K and centrifuged at 4000 rpm per 1 min.

Samples of sewage scum were taken from different urban WWTPs: Polignano a Mare (12000 PE), Putignano (12000 PE), Bisceglie (50000 PE), Barletta (90000 PE), Andria (100000 PE) and Bari West (242000 PE). These different samples were first characterised in terms of total solids, ashes and total lipids. Table 1 shows their respective starting compositions. A part different water contents, and consequential TS values, the amounts of lipids calculated respect to TS values, always range between 35 and 50%. Even ashes and soaps contents seem to be ranged into a very strict interval respectively between 5 and 7% respect to TS and 30e40% respect to overall lipids. In any case, FFAs remain the most present component into total lipids, whereas glycerides (mono-, di- and tri-) were completely absent. These very redundant compositions of sewage scum collected from the mentioned WWTPs were in good agreement with compositions of samples of scum taken from WWTPs located in other different

58

L. di Bitonto et al. / Renewable Energy 90 (2016) 55e61

Fig. 2. Overall scheme of process of sequential batch reactors using a counter current flow of reactants, for the conversion of lipids of sewage scum into FAMEs.

Table 1 Summary results on characterisation and recoverability of lipids through centrifugation of the heated sewage scum. Entry

WWTPs

Total solids (%wt)

Ashes (%wt)

Total Lipidsc (%wt)

Soapsd (%wt)

Condition of Centrifugation Speed, T, t (rpm, K, min)

Recoverabilitye (%)

1 2 3 4 5 6 7 8 9 10 11

Polignano

35.2 ± 0.5

2.2

15.0 ± 0.1

28 ± 1

Polignanoa Polignanob Putignano Bisceglie Barletta Andria Bari West

17.5 8.8 19.3 16.6 22.9 17.3 22.4

1.1 0.55 1.7 1.2 1.8 1.4 1.7

7.5 3.75 8.4 ± 0.1 6.3 ± 0.3 9.4 ± 0.2 6.2 ± 0.1 11.2 ± 0.2

28 28 34 27 33 35 30

3000, 4000, 4000, 4000, 4000, 4000, 4000, 4000, 4000, 4000, 4000,

57.1 ± 0.6 98.7 ± 1.2 no separation 98.0 ± 1.1 93.5 ± 0.5 92.5 ± 0.5 94.3 ± 0.6 92.4 ± 0.7 95.2 ± 0.6 94.4 ± 0.8 95.4 ± 0.8

± ± ± ± ±

0.5 0.5 0.4 0.2 0.3

± ± ± ± ± ± ±

1 1 1 1 1 1 1

353, 353, 343, 353, 353, 353, 353, 353, 353, 353, 353,

2 2 2 1 1 1 1 1 1 1 1

WWTPs ¼ wastewater treatment plants, T ¼ temperature, t ¼ time, %wt ¼ weight percentage. a Original sample uptaken from WWTP of Polignano diluted with tap water 1:1. b Sample “a” diluted with tap water 1:1. c Total lipids expressed as percentage referred to starting wet sample. d Percentage referred to lipids. e Value calculated by the ratio of lipid amount separated from centrifugation and total lipid.

countries: from Xi'an (China) [7], to Traiguen (Chile) [8] and St Paul (Minnesota, USA) [9] a very large presence of FFAs is always detected. In order to optimise the best separating conditions of lipids through centrifugation of the heated scum, the effect of temperature, time and speed of centrifugation were initially evaluated on samples taken from the WWTP of Polignano, for which lower amount of water content was found. Temperature and speed of centrifugation had a positive effect on separation. Slower was the speed of centrifugation (please see entries 1 and 2), worse was recoverability of lipids from solids. Fixing the temperature at 353 K and the time of centrifugation at 2 min, recovered lipids decreased from 98% at 4000 rpm to 57% at 3000 rpm. On the other hand, operative temperature needed to be set to 353 K at least, since already working at 343 K, separation of lipids was not appreciated (please see entries 2 and 3). Finally, time of centrifugation represented the variable with the lowest influence: significant changes were not detected by decreasing from 2 to 1 min (entries 2 and 4). At the end, centrifugation at 4000 rpm per 1 min of heated sewage scum at 353 K, resulted the best operative conditions for separating lipids. Secondly, the effect of water content was determined on the separation process working on sewage scum taken from Polignano, by simply adding a known amount of tap-water on the raw sample. In Table 1 (entries 4, 5 and 6) are reported the recovery of lipids into the different starting wet samples. Actually, the slight decrease of recovery of lipids verified with diluting the starting samples was

mainly due to a physical difficulty of efficiently pipetting away lipid phase at the end of centrifugation, because also in these cases, separation between grease and the aqueous layer was neat. These optimised conditions were then applied to sewage scum taken from other different WWTPs. Very interestingly, lipids extracted represented always 93e98% of the overall oil previously determined on the respective samples, demonstrating the good efficacy of the proposed technique. Then, the FAs profile were determined from the recovered oily phases, obtaining results reported into Table 2. Table 2 shows how similar the FAs composition looked like, even though starting scums were sampled from very different places. Such a similar composition represents a key positive point on the physical behaviour of the final product, which very presumably, still remain the same independently from where it is produced. All these analogies related to starting compositions, make sewage scum a worldwide diffuse potential feedstock for producing biodiesel, whose chemicalephysical properties should be very similar. Once centrifuged, the upper oily phase was kept at 378 K for 24 h, appreciating a very limited loss of weight: lower than 1%wt. Determinations of free acidity and soaps carried out on these samples allow the determination of the exact amount of formic acid needed for converting soaps into the relevant acid to be defined. When stoichiometric amount of formic acid respect to soaps was added, precipitation of solids (analysed as Ca(HCOO)2 [15]) was obtained, centrifuged at 333 K and easily separated from the

L. di Bitonto et al. / Renewable Energy 90 (2016) 55e61

59

Table 2 FFAs composition of the lipid phase extracted from sewage scum. Fatty acids (FAs) Lauric Acid (C12:0) Myristic Acid (C14:0) Palmitic Acid (C16:0) Palmitoleic Acid (C16:1) Stearic Acid (C18:0) Oleic Acid (C18:1) Linoleic Acid (C18:2) AMW FAs (g/mole)

Polignano (%wt)

Putignano (%wt)

Bisceglie (%wt)

Barletta (%wt)

Andria (%wt)

Bari West [7] (%wt)

0.5 4.8 34.3 3.2 8.7 48.5

1.8 5.1 30.6 3.2 7 52.3

1.1 4.4 30.8 2.2 10.2 51.3

1 4.2 31.5 2.9 7.4 49.5

1.3 4.6 30.7 2.7 12.5 48.2

1 2.4 24.4 2.2 7.9 62.1

269.8

269.5

270.8

260.6

270.4

273.6

Xi'an (%wt) 0 4 26.5 4 7.5 41 17 272.1

St Paul [9] (%wt) 0 5.6 32.5 2.3 15.6 41.3 263.0

AMW ¼ Average molecular weight, %wt ¼ weight percentage.

activated oily phase. Recovery of the activated oily phase was always very efficient, because at the end 80e85% of starting oil were separated at least, whose FFAs components was between 82 and 85%wt.

3.2. Conversion of activated lipids into FAMEs When activated oily phase was reacted with methanol and AlCl3$6H2O, under a molar ratio of 1:10:0.02 for 2 h at 345 K, an overall conversion of starting FFAs into FAMEs of 93.4% was obtained. The final conversion of FFAs into FAMEs did not exceed 94%, even when prolonged reaction time of 12 h was adopted, to confirm a thermodynamic cap. On the other hand, a very convenient separation of the reaction mixture at the end of the direct esterification process was verified [10]: most of unreacted methanol, with the water produced during the reaction and with most of the catalyst used, resulted completely separated in an upper phase on the oily layer, in which most of FAMEs (>95%) were concentrated. Residual FFAs resulted equally distributed in both phases. This very favourable separation allowed a very simple recoverability of the catalyst into the methanol phase: without isolating it, its successive reuse in new cycles of reactivity was plenty verified for more than three times. For this reason, a new sequence of three batch-reactors in a counter-current configuration of reactants was applied with the aim of fully exploiting methanol and catalysts amounts to push the conversion of FFAs into FAMEs to values bigger than the above mentioned 94%. Fig. 2 shows the scheme of process, in which three reactors (R1, R2 and R3) were run. At the end of each cycle of 2 h at 345 K, agitation was stopped and the three solutions were left to decant by determining the separation of six different phases (O1 and M3 from R1, O2 and M2 from R2 and O3 and M1 from R3). Fresh oil was fed in Reactor 1 together with M2, which was the methanol phase already used for two previous cycles. Whereas fresh methanol and catalyst were charged into reactor R3 with O2. M3 and O3 represented respectively the exhausted methanol phase and the oily product recovered respectively from R1 and R3, after having been reacted for three cycles. With this configuration of reactors, a gradient of FFAs to be converted into the relevant FAMEs were always maintained going from R1 to R3, allowing at the end, a more efficient final conversion to be achievable. On the contrary, as for the methanol phase, an increase of water content was verified going from R3 to R1. Fig. 3a clearly shows the amounts of O3 and M3 recovered after 14 cycles of reaction. Whereas Fig. 3b and c show profiles of concentration of the two collected phases along the same experiments. After a first slight change of water content into the methanol exhausted phase (M3) and a slight increase of FAMEs into the final oil (O3), the final compositions and the relative amount of each phases were subjected to an evident stabilisation. Oscillations rapidly converged into a very narrow interval of variation (less than

3%) already after 3 cycles. Interestingly, the above mentioned trend of the separation described for a single batch reaction, resulted perfectly confirmed: 98% of water produced into the reaction was collected into the exhausted methanol phase (M3), as well as 97% of aluminium catalyst (Fig. 3c). Whereas, on the other side, most of FAMEs (>95%) were accumulated into O3 (Fig. 3b) with a residual water content of 0.2%wt. Once that all three reactors were filled, and the stationary phase was reached, at the end of each cycle, fresh oil and methanol were charged, while M3 and O3 were recovered with a final overall conversion of FFAs into FAMEs higher than 99%. At the end, methanol should be evaporated from O3, whereas purification of FAMEs should be obtained through distillation under vacuum of residue [10]. 3.3. Digestibility of residues of centrifugation The overall sustainability of the process can be better appreciated by considering the fate of residues generated from centrifugation. Besides the separation of lipids and an aqueous middle phase (Fig. 1) a wet solid could be recovered, whose main organic components resulted cellulose (25e30%), lignin (30e35%) and proteins (20e25%). The use of a centrifugation without the addition of any organic solvents, made this residue actually ready to be digested anaerobically [16]. Specifically, if used in anaerobic digestion, it would produce 47.5 Standard Temperature Pressure (STP) m3 bio-methane per m3 of starting sewage scum [17]. This volume of bio-methane would be largely enough to supply the required energy of the overall process (about 5 times the energy needed for heating 1 m3 of starting sewage scum from 283 to 353 K). In this way, the starting sewage scum should be completely consumed (unless lignin and insoluble ashes), by effectively producing biodiesel. 3.4. Potential of technology Having already verified a very similar starting composition of sewage scum sampled in different countries, the capability of fully exploiting the potential of sewage scum to produce biodiesel should not have big technological limitations. Really interestingly, even the annual amount of lipids (about 300 g) recoverable from sewage scum per PE estimated for an Italian little WWTP (Polignano) resulted very similar to that evaluated for a big plant of US (Metro Plant of St Paul in Minnesota). Apparently, there are only two critical aspects that could subdue the development of this technology. The first is correlated to the dimension of WWTPs and consequentially to the respective overall volume of sewage scum annually collected. In fact, if for big WWTPs the large amount of sewage scum could justify the realisation of the complete transformation-chain, for little plants, the reduced volume of scum could not justify any investment in this direction. Secondly, in order to apply this technology, it is supposed a

60

L. di Bitonto et al. / Renewable Energy 90 (2016) 55e61

realised. In any case, a two-step configuration could help in overcoming the limitations; by simply separating competences into two distinguished parts: the first phase of centrifugation of scums should be of competence of WWTPs, whereas the final transformation of the separated raw lipids should be addressed to biodiesel companies. In this way, even little WWTPs should find profitable to process sewage scum. The solvent-less extraction described in this work limits the extraction process to a simple centrifugation, which is an operation already used in WWTPs for dewatering sludge, so that they could sell the raw lipid to a final receiver, instead of paying the disposal of sewage scum as special waste. On the other hand, biodiesel producers should have new sustainable feedstock at a convenient price for producing FAMEs. According to the above mentioned estimation of lipids from scums produced per PE, the annual potential productivity of biodiesel from this waste in Europe should be accounted to around 0.15 Mton, corresponding to 1.5% of the present overall continental production. This productivity should be counted doubled, according new European legislative criteria, for being obtained from a waste oil fully respecting sustainability criteria. 4. Conclusions In this work, applicability of a solvent-less separating procedure of lipids from wet sewage scum was positively tested. Through a simple centrifugation (4000 rpm, 1 min) of heated raw sewage scum (355 K), a complete recovery of lipids (93e98%) could be determined. Even sewage scum taken from very little WWTPs could be efficiently used for these purposes. The analogous composition determined on different samples make sewage scum a worldwide diffuse potential feedstock for producing biodiesel. Besides the very sustainable recovery of lipid, a full optimisation of conversion of FFAs into FAMEs was achieved. By adopting three sequential batch reactors configured in a counter current flux of reactants, the final conversion of starting FFAs was improved from the thermodynamic limit of a single batch (93.7%) upto e99%, leaving the FFAs:methanol:catalyst molar ratio unchanged (1:10:0.02), as well as reaction conditions (345 K, 2 h per cycle). After this reactive treatment, residual methanol and water were boiled away from the oily product, before to carry out the final distillation under vacuum of pure biodiesel. The final distillate resulted conform to EN14214 standard specifications [10]. Although such purification processes increase the overall production cost of biodiesel, it still remains absolutely competitive with conventional productions, which use refined oils [18,19]. At the end, besides the obtainment of biodiesel, a complete exhaustion of the waste could be realised through anaerobic digestion of the residual solid waste. Residual cellulose and proteins would produce enough amount of bio-methane to sustain the overall process in terms of electric and heating energy demand (47.5 STP m3 bio-methane per m3 of starting sewage scum).

Fig. 3. a) Masses of O3 and M3 recovered for 14 cycles of reaction; b) profiles of percentage composition of methanol, water, FAMEs and FFAs revealed in O3 throughout 14 cycles of reaction; c) profiles of percentage composition of methanol, water, Fatty Acid Methyl Esters (FAMEs), Free Fatty Acids (FFAs) and AlCl3$6H2O determined in M3 throughout 14 cycles of reaction. R ¼ Reactor; M1, M2 ¼ methanol phases separated by decanting reaction mixtures obtained by R3 and R2 respectively; O1, O2 ¼ oily phases separated by decanting reaction mixtures obtained by R1 and R2 respectively; O3 ¼ oily product, M3 ¼ exhausted methanol.

fundamental change of conception of WWTPs into sort of “chemical industries”: the need of high qualified personnel, new machineries and the use of methanol could represent too drastic changes to be

Acknowledgement This study was financially supported by CNR through “Bioraffineria di terza generazione integrata con il territorio e biocombustibili” Progetto Premiale. The authors would like to acknowledge Acquedotto Pugliese SpA (AqP) for the kind collaboration. References [1] A. Ajanovic, Renewable fuels e A comparative assessment from economic, energetic and ecological point-of-view up to 2050 in EU-countries, Renew.

L. di Bitonto et al. / Renewable Energy 90 (2016) 55e61 Energy 60 (2013) 733e738. [2] Directive 2009/28/EC of the European Parliament and of the Council of 23 April 2009 on the promotion of the use of energy from renewable sources and amending and subsequently repealing Directives 2001/77/EC and 2003/30/EC. [3] L. Martín-Gonz alez, R. Castro, M.A. Pereira, M.M. Alves, X. Font, T. Vicent, Thermophilic co-digestion of organic fraction of municipal solid wastes with FOG wastes from a sewage treatment plant: reactor performance and microbial community monitoring, Bioresour. Technol. 102 (2011) 4734e4741. [4] M.J. Montefrio, T. Xinwen, J.P. Obbard, Recovery and pre-treatment of fats, oil and grease from grease interceptors for biodiesel production, Appl. Energy 87 (2010) 3155e3161. [5] C.J. Stacy, C.A. Melick, R.A. Cairncross, Esterification of free fatty acids to fatty acid alkyl esters in a bubble column reactor for use as biodiesel, Fuel Process. Technol. 124 (2014) 70e77. [6] C. Pastore, A. Lopez, G. Mascolo, Efficient conversion of brown grease produced by municipal wastewater treatment plant into biofuel using aluminium chloride hexahydrate under very mild conditions, Bioresour. Technol. 155 (2014) 91e97. [7] W. Yi, F. Sha, B. Xiaojuan, Z. Jingchan, X. Siqing, Scum sludge as a potential feedstock for biodiesel production from wastewater treatment plants, Waste Manag. 47 (2016) 91e97. [8] N. Sangaletti-Gerhard, M. Cea, V. Risco, R. Navia, In situ biodiesel production from greasy sewage sludge using acid and enzymatic catalysts, Bioresour. Technol. 179 (2015) 63e70. [9] C.H. Bi, M. Min, Y. Nie, Q.L. Xie, Q. Lu, X.Y. Deng, E. Anderson, D. Li, P. Chen, R. Ruan, Process development for scum to biodiesel conversion, Bioresour. Technol. 185 (2015) 185e193.

61

[10] C. Pastore, E. Barca, G. Del Moro, A. Lopez, G. Mininni, G. Mascolo, Recoverable and reusable aluminium solvated species used as a homogeneous catalyst for biodiesel production from brown grease, Appl. Catal. A Gen. 501 (2015) 48e55. [11] O.H. Lowry, N.J. Rosebrough, A.L. Farr, R.J. Randall, Protein measurement with the folin phenol reagent, J. Biol. Chem. 193 (1951) 265e275. [12] ISO 11465, Soil Quality e Determination of Dry Matter and Water Content on a Mass Basis e Gravimetric Method, 1993. [13] NREL TP-510-42618, Determination of Structural Carbohydrates and Lignin in Biomass. [14] C. Pastore, A. Lopez, V. Lotito, G. Mascolo, Biodiesel from dewatered wastewater sludge: A two-step process for a more advantageous production, Chemosphere 92 (2013) 667e673. [15] C. Pastore, M. Pagano, A. Lopez, G. Mininni, G. Mascolo, Fat, oil and grease waste from municipal wastewater: characterization, activation and sustainable conversion into biofuel, Water Sci. Technol. 71 (2015) 1151e1157. [16] M. Olkiewicz, M.P. Caporgno, A. Fortuny, F. Stüber, A. Fabregat, J. Font, C. Bengoa, Direct liquideliquid extraction of lipid from municipal sewage sludge for biodiesel production, Fuel Process. Technol. 128 (2014) 331e338. [17] I. Angelidaki, W. Sanders, Assessment of the anaerobic biodegradability of macropollutants, Rev. Environ. Sci. Biotechnol. 3 (2004) 117e129. [18] M.R. Avhad, J.M. Marchetti, A review on recent advancement in catalytic materials for biodiesel production, Renew. Sust. Energy Rev. 50 (2015) 696e718. [19] Y.D. You, J.L. Shie, C.Y. Chang, S.H. Huang, C.Y. Pai, Y.H. Yu, C.H. Chang, Economic cost analysis of biodiesel production: case in soybean oil, Energy Fuel 22 (2008) 182e189.