Volatile fatty acids platform from thermally hydrolysed secondary sewage sludge enhanced through recovered micronutrients from digested sludge

Water Research 100 (2016) 267e276

Contents lists available at ScienceDirect

Water Research journal homepage: www.elsevier.com/locate/watres

Volatile fatty acids platform from thermally hydrolysed secondary sewage sludge enhanced through recovered micronutrients from digested sludge Philemon J. Kumi a, *, Adam Henley a, Achame Shana b, Victoria Wilson c, Sandra R. Esteves a, ** a

Wales Centre of Excellence for Anaerobic Digestion, Sustainable Environment Research Centre (SERC), University of South Wales, Pontypridd, MidGlamorgan, CF37 1DL, UK Thames Water Limited, Reading, Berkshire, RG1 8DB, UK c ^ Cymru Welsh Water, Nelson, Treharris, Mid-Glamorgan, CF46 6LY, UK Dwr b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 10 March 2016 Received in revised form 6 May 2016 Accepted 7 May 2016 Available online 12 May 2016

The extracellular polymeric substances and microbial cytoplasmic contents seem to hold inorganic ions and organic products, such as proteins and carbohydrates that are of critical importance for the metabolism of hydrolytic and acidogenic anaerobic microorganisms. The addition of soluble microbially recovered nutrients from thermally treated digestate sludge, for the fermentation of thermally hydrolysed waste activated sludge, resulted in higher volatile fatty acids yields (VFAs). The yield of VFAs obtained from the recovered microbial nutrients was 27% higher than the no micronutrients control, and comparable to the yield obtained using a micronutrients commercial recipe. In addition, the use of a low pH resulting from a high sucrose dose to select spore forming acidogenic bacteria was effective for VFA production, and yielded 20% higher VFAs than without the pH shock and this associated with the addition of recovered microbial nutrients would overcome the need to thermally pre-treat the inoculum. © 2016 Elsevier Ltd. All rights reserved.

Keywords: Waste activated sludge Hydrolytic-acidogenic fermentation Micronutrients Volatile fatty acids Inoculum selection

1. Introduction A significant focus of research into anaerobic digestion process (AD) in recent times, has been the enhancement of the acidogenic process, for the production of volatile fatty acids (VFAs), as essential precursors for the second-stage biomethanation process (Massanet-Nicolau et al., 2013), or as substrate for polyhydroxyalkanoate (PHA) production (Kedia et al., 2014), or for bioelectrochemical systems (BES) (Guwy et al., 2011). A variety of feedstocks, including, foodwaste, organic fractions of municipal solid waste (OFMSW), dairy wastewater and olive oil mill waste, have been evaluated for the production of VFAs (Lee et al., 2014). Some of the process conditions that have critically been evaluated in the optimisation of the production of VFA include: organic loading rate (OLR) (Kyazze et al., 2006),

* Corresponding author. ** Corresponding author. E-mail addresses: [email protected] (P.J. Kumi), sandra.esteves@ southwales.ac.uk (S.R. Esteves). http://dx.doi.org/10.1016/j.watres.2016.05.030 0043-1354/© 2016 Elsevier Ltd. All rights reserved.

temperature (Yuan et al., 2009), pH (Massanet-Nicolau et al., 2008), hydraulic retention time (HRT) (Massanet-Nicolau et al., 2009) and additives such as surfactants and enzymes (Lee et al., 2014). It is important to note however that most of the previous research related to VFA production was conducted for optimisation of biohydrogen production, and not necessarily to investigate or optimise VFA production. In particular for sewage sludge, studies related to VFA production have been limited to a couple of lab-scale (semi) continuous fermentation studies (Morgan-Sagastume et al., 2011; Maharaj and Elefsiniotis, 2001) and very few full scale applications (Shana et al., 2003). The selection of inoculum source, such as activated sludge, aerobic compost, thermally treated anaerobically digested sludge and soil, have also been identified as an important step in the optimisation of the hydrolytic-acidogenic fermentation process for VFA production (Hawkes et al., 2002). Thermally treated digested sewage sludge is often selected as inoculum for VFA and biohydrogen production, in that, the thermal treatment helps to eliminate the presence of VFA and hydrogen utilising microorganisms, leaving spore-forming bacteria species (Hawkes et al., 2002;

268

P.J. Kumi et al. / Water Research 100 (2016) 267e276

Wang et al., 2003). As explained by Hawkes et al. (2002) bacterial species, such as Bacillus and Clostridium, undergo sporulation under unfavourable or stressful environmental conditions such as rising temperatures, low or high pH and depletion of nutrients. Nonspore forming microorganisms on the other hand, undergo cell lyses under unfavourable environmental conditions. As observed by Massanet-Nicolau et al. (2008), spore formers effectively undergo sporulation by heating mixed microflora at 110
C for 20 min whilst non-spore formers undergo cell lyses. The cytoplasmic content of non-spore forming microbes is therefore expected to be released to solution after thermal treatment. According to Feeherry et al. (1986), even spore-forming bacteria undergo injury and eventually to cell lyses at 124
C within 1 min; 62 min is however required to achieve same results at 112.8
C. Thermal treatment for the selection of acidogenic microbes is also expected to cause the dissolution of extracellular polymeric substances (EPS) (a complex high-molecular-weight mixture of polymers) produced by the microorganisms that surrounds the cell wall, which also contain inorganic ions. Soluble microbial products such as nucleic acids, uronic acids, proteins, lipids and carbohydrates, also become available after thermal treatment of digested sludge (Sheng et al., 2010). Several studies have investigated the influence of various micronutrients on AD (Feng et al., 2010). There is however little knowledge about the influence of micronutrients on hydrolytic-acidogenic fermentation. This study aims to evaluate the importance of selection of inoculum and the use of micronutrients in VFA production from thermally hydrolysed secondary sewage sludge or waste activated sludge (WAS). The study also explores the use of soluble microbial nutrients (potentially in the EPS and soluble microbial products) after thermal treatment of digested sludge as micronutrients source to enhance VFA production. 2. Materials and methods The acidogenic fermentation was carried out in 2.5 L reactors with working volume of 2-L. The reactors were kept in a shaking incubator equipped with temperature controllers, which maintained the operation temperature at 37 ± 1
C. The incubator provided a constant agitation at 110 rpm. The inoculum was obtained from Welsh Water at the Cog Moor Wastewater plant; a conventional AD plant treating largely secondary sewage sludge without any pre-treatment. The feedstock for the fermentation process was thermally hydrolysed waste activated sludge (TH-WAS), obtained from Welsh Water treatment plant in Cardiff. The thermal hydrolysis pre-treatment is carried out at 165
C and 6 bars for 30 min. The TH-WAS feedstock was collected at 2 weeks interval and stored in a refrigerator at 4
C. All the reactors were operated at semicontinuous mode (once a day feeding during the week days), at an OLR of 25.54 gVS L1 d1. The reactors were operated at HRT of 2.8 days. The pH of the reactor content during the fermentation process was not controlled in the study. Samples were taken periodically for VFA measurement. The TH-WAS feedstocks for some of the reactors were supplemented with a tailor-made commercial micronutrients (CM) (Table 5 shows the concentration trace minerals in the CM added). The CM was added to the TH-WAS substrate and stored at 4
C for later use. Two runs of the same experiments were performed with the repeat taking place one month apart. 2.1. Inoculum preparation and micronutrients recovery from digested sludge The digested sludge was sieved through a 2 mm mesh to remove particulate matter. The sieved sludge was either used directly or heated at 110
C for 20 min.

A portion of the thermally treated sludge was then centrifuged at 3000 rpm for 20 min, after being allowed to cool to ambient temperature. The supernatant was decanted, and then filtered through a 0.2 mm pore size hollow-fibre membrane module (Tianjin Motian Membrane Eng. and Tech. Co. Ltd.), to remove any particulate matter as well as bacterial cells. The filtrate was heat-treated at 110
C, and then stored at 4
C for later use as micronutrient source (Recovered microbial nutrients, RMN). 2.2. Experimental set-up The study included six experimental set-ups involving inoculum pre-treatments and micronutrients addition, and their controls, as shown in Table 1. The start-up of all the NI-reactors (NI, NI-CM and NI-RMN) involved a high loading of sucrose (a concentration of 30 g/l) causing a pH reduction and therefore making the environmental conditions more favourable for the acidogens, the important microbial population for the synthesis of VFAs. After the initial feeding with sucrose, the NI reactors were left for 56 h without feed, at which time the pH of the reactor content reduced from an initial 7.35 to 4.05. The reactors were then fed with the TH-WAS substrate. The TH-WAS substrate was fed to the reactors in a semi-continuous mode by withdrawing (for analysis) and feeding the same amount daily. The N0 reactor used as control for the NI-reactors involved the use of untreated inoculum with no sucrose dosing. A schematic illustrating the experimental desiagn is shown in Fig. 1. 2.3. Analytical methods The VFA concentrations in the samples were determined as described by Cruwys et al. (2002) using a head space gas chromatograph equipped with a flame ionisation detector (Perkin Elmer, UK). Soluble COD was measured using the Hach Lanch COD kit. Soluble carbohydrate was determined in using the phenolsulphuric assay (Dubois et al., 1956). The type and quantity of elements in the soluble fraction of the sludge, CM and RMN solutions were determined using inductively coupled plasma optical emission spectrometry (ICP-OES) as described by Fassel and Knlseley (1974). 3. Results and discussions The different conditions of the acidogenic fermentation were repeated twice at one month interval. As shown in Table 2, the VFA yield obtained from Run 2 was generally higher than the yield obtained in Run 1. The difference in the two runs was in the concentration of the feedstocks, due to full scale variation in sludge concentration, which after preparation in the lab were 7.28 ± 0.43 g L1 TS (59.03± 1.27% VS) in Run 1, compared to 8.23 ± 0.17 g L1 TS (62.07 ± 0.35) in Run 2. The OLR in Run 2 was also higher than in Run 1, and that potentially resulted in the difference in the yield of VFAs. As shown in Table 2, the patterns of increase of VFA with the different conditions in the two runs were very similar. The major difference in the two runs was the VFA yield from the N0 reactor, which was nearly 72% higher in Run 2, as compared to Run 1. The lack of selecting specific microbial strains for acidogenic fermentation makes the digester more susceptible to process instability during start-up of the fermentation process. It can be deduced that acidogenic fermentation in the N0 reactor, was more active in Run 2 compared to Run 1 and that could have resulted from the higher loading in Run 2, which potentially contributed to an improved acidogenic promoting condition with a reduced pH. The discussion in this study was based on data analysis from Run 2. The

P.J. Kumi et al. / Water Research 100 (2016) 267e276

269

Table 1 Experimental conditions. Reactors

Inoculum

Initial sucrose dosingc

Micronutrients

N0 NI HI NI-CM HI-CM NI-RMN

Untreateda Untreated Heat treatedb Untreated Heat treated Untreated

Nil Yes Nil Yes Nil Yes

Nil Nil Nil Commercial micronutrients Commercial micronutrients Recovered microbial nutrients

a b c

No heat treatment. Heat treatment at 110
C for 20 min. Sucrose dose at 30 g/l once, followed by 56 h of fasting.

determination of the average yield of VFA was carried out after 5 HRTs (day 14e31) from the initial TH-WAS feeding. 3.1. Selection of optimal VFA producing inoculum The sucrose (30 g/l) fermentation in the NI reactors, followed by 56 h of fasting, prior to the TH-WAS feeding, resulted in an average inoculum pH of 4.00 ± 0.026 (Fig. 2). The initial addition of the THWAS substrate increased the pH to 4.92, 4.96, 5.05 in the NI, NI-CM, NI-RMN, respectively. The average pH of the HI inoculum was 7.30 ± 0.02, reducing to 7.28 and 7.20 in the HI and HI-CM respectively, at the start of the TH-WAS feeding (Fig. 2). The pH of the N0 inoculum was 7.05. After the initial TH-WAS feeding, the pH

increased to 7.32. During the steady-state of substrate fermentation (day 14e31), the pH in N0 reactor reduced to 6.75 ± 0.06, compared to 5.96 ± 0.14 pH observed in the NI reactor. The results of an independent t-test (Table 3b) showed that the mean difference between pH in NI and N0 reactors were statistically significant, t ¼ 7.22, p ¼ 0.000. The effect size, Cohen’s d (4.169) > 0.80 was large, indicating that the distribution of the mean values of the pH in the two reactors were very different from one another. As shown in Table 3b, the difference in the mean values of pH between the NI reactors (NI, NI-CM and NI-RMN) and HI reactors (HI, HI-CM and HI-RMN) were statistically significant from each other, p values < 0.05. The effect size between the NI and HI reactors were all very large, indicating that the distribution of the mean values of

Semi-conƟnuous reactors operated on TH-WAS CM HI-CM Reactor

HI Reactor

Centrifuging

Thermal Treatment at 110oC

Supernatant

Digested sewage sludge

0.2 μm Filtration

30g/l sucrose shock

RM

NI-RM Reactor

CM NI Reactor

N0 Reactor

-1

16.56 gVFACOD L

NI-CM Reactor

-1

19.84 gVFACOD L

-1

25.11 gVFACOD L

-1

24.03 gVFACOD L

NB: TH-WAS – Thermally hydrolysed waste acƟvated sludge; RMN – recovered microbial nutrients; CM – commercial micronutrients

Fig. 1. Schematic illustration of experimental setups.

270

P.J. Kumi et al. / Water Research 100 (2016) 267e276

Table 2 Characteristics of TH-WAS feedstock and resultant VFA yield.

Parameters HRT (days) Temperature (
C) OLR (gVS L1 d1) TH-WAS substrate TS (%) VS (%) VFA (gVFACOD L1) NH4 (g L1) pH VFA yield (gVFACOD L1) N0 NI HI HI-CM NI-CM NI-RMN VFA yield increase (%) NI vs N0 HI vs NI HI-CM vs HI NI-CM vs NI HI-CM vs NI-CM NI-RMN vs NI NI-RMN vs HI NI-RMN vs NI-CM NI-RMN vs HI-CM

Run 1

Run 2

2.8 37 ± 1 21.49

2.8 37 ± 1 25.54

7.28 ± 0.43 (n: 3) 59.03 ± 1.27 (n:3) 6.35 ± 0.95 (n:2) 2.31 ± 0.15 (n:2) 6.6 (n:1)

8.23 ± 0.17 (n:3) 62.07 ± 0.35 (n:3) 5.86 ± 0.23 (n:3) 1.93 ± 0.08 (n:2) 5.8 (n:1)

9.54 ± 0.54 (n:6) 16.38 ± 0.88 (n:7) 18.85 ± 0.96 (n:7) 19.28 ± 0.54 (n:9) 19.29 ± 0.43 (n:15) 20.26 ± 0.92 (n:14)

16.56 19.84 22.98 24.68 24.03 25.11

72 15 2 18 0 24 7 5 5

20 16 7 21 3 27 9 4 2

± ± ± ± ± ±

0.46 0.28 0.33 0.27 0.32 0.27

(n:12) (n:14) (n:12) (n:12) (n:12) (n:12)

NB: VFA yields in Run 2 is average VFA measurement during fermentation steady state (day 1431) as shown in Figs. 3 and 5; n e sample size.

the pH in the two groups of reactors were very different from each another. The results depicted different environmental conditions in the reactors based on the kind of inoculum pretreatment. The low initial pH after feeding in the NI reactors (around pH 5, as shown in Fig. 2) favoured the hydrolytic bacteria, whose activity produced organic acids, which kept the pH during the continuous feeding of the TH-WAS substrate. The relatively low pH in the NI and high pH in the N0 at the beginning and during stable fermentation of the TH-WAS substrate, in this study, suggested that pH was a significant factor in the microbial activity resulting in the higher yield of VFAs. The use of pH to repress methanogenic activity has been

land (2006), who similarly, observed no investigated by Zhu and Be methanogenic activity when the starting pH was below 5.0. As they suggested, a low pH apart from selecting acidogenic bacteria, may also have a fatal effect on hydrogen production; hence the negligible yield of hydrogen in the NI compared to the N0. It could be suggested that the low pH shifted the metabolic activity towards VFA production. The cumulative hydrogen production in all the reactors ranged between 0.29 and 4.78 L H2 kgVS1 added (Table 3a). The relatively low pH due to the high loading of sucrose, and resultant high VFAs, at the start of the NI experiment, coupled with the low HRT and high OLR of the sewage sludge in the fermentation process was sufficient to create environmental conditions that were not favourable to the methanogens, thereby repressing methanogenic activity; indicated by the lack of methane production in the NI reactors (Table 3a). Similarly, the heat treatment of the inoculum in the HI setups, together with low HRT and high OLR, was also effective in the repression of the methanogenic activity, hence the lack of methane production in the HI and HI-CM reactors. In the N0 control experiment, however, an average methane concentration of 11.30 ± 1.48% was observed, indicating the lack of complete repression of the methanogenic activity (Table 3a). The yield of hydrogen from the NI and NI-CM reactors were negligible when compared to the yield obtained from the N0, HI and HI-CM reactors. Although the hydrogen yield from the NI reactor (as shown in Table 3a) was significantly lower than that obtained from the N0 reactor, the higher yield of VFA in the NI reactors, indicated that microbial activity was directed towards the production of VFA, rather than hydrogen. As shown in Table 3a, the NI reactor produced a total VFA of 19.84 ± 0.28 g VFACOD L1, compared to the N0 (control), which resulted in a total VFA production of 16.56 ± 0.46 g VFACOD L1, representing a VFA yield increase of approximately 20%. The results of an independent t-test (Table 3b) showed that the mean difference between VFA yields from NI and N0 reactors were statistically significant, t ¼ 6.28, p ¼ 0.000. The yield of VFA from the HI reactor was 22.98 ± 0.35 g VFACOD L1, which was approximately 16% higher than the yield obtained from the NI reactor. The mean difference between VFA yields from HI and NI reactors were also statistically significant, t ¼ 7.36, p ¼ 0.000. The difference in the distribution of the mean values (effect size) of VFA yields between

8.00 7.00

6.00

pH

5.00 4.00 3.00 2.00 1.00

0.00 N0

NI

HI

HI-CM

NI-CM

NI-RMN

pH of inoculum pH aŌer iniƟal TH-WAS substrate addiƟon pH during steady state of TH-WAS substrate fermentaƟon Note: pH before the addition of the TH-WAS substrate (after the sucrose fermentation) in the NI, NI-CM and NI-RMN was about pH 4.0, pH during steady state of fermentation was an average of two pH measurement per week from day 14-31 Fig. 2. pH change in reactors at start of TH-WAS substrate fermentation and at steady-state fermentation conditions.

P.J. Kumi et al. / Water Research 100 (2016) 267e276

271

Table 3a Products of the acidogenic fermentation process.

Cumulative H2 (L H2/kg VS) Average methane (%) Acetic (g VFACOD/gHAc) Propionic (g VFACOD/g HPr) iso-butyric (g VFACOD/g HBu) n-Butyric (g VFACOD/g HBu) iso-valeric (g VFACOD/g HVa) n-valeric (g VFACOD/g HVa) Mean tVFA (g VFACOD L1)a pHa

N0 reactor

NI reactor

HI reactor

HI-CM reactor

NI-CM reactor

NI-RMN reactor

1.39 11.3 ± 1.48 (n:4) 5.31 ± 0.20 2.56 ± 0.09 1.22 ± 0.06 3.45 ± 0.24 1.88 ± 0.09 2.12 ± 0.17 16.56 ± 0.46 (n ¼ 12) 6.75 ± 0.14 (n ¼ 6)

0.13 n.d. 6.91 ± 0.18 2.90 ± 0.05 1.45 ± 0.04 4.00 ± 0.10 2.73 ± 0.05 1.65 ± 0.10 19.84 ± 0.28 (n ¼ 14) 5.96 ± 0.14 (n ¼ 6)

4.53 n.d. 8.48 ± 0.19 3.51 ± 0.09 1.69 ± 0.03 4.71 ± 0.10 2.93 ± 0.08 1.74 ± 0.08 22.98 ± 0.33 (n ¼ 12) 6.31 ± 0.08 (n ¼ 6)

4.78 n.d. 9.66 ± 0.18 3.98 ± 0.09 1.81 ± 0.03 4.58 ± 0.04 2.75 ± 0.06 1.99 ± 0.02 24.68 ± 0.27 (n ¼ 12) 6.14 ± 0.07 (n ¼ 6)

0.32 n.d. 9.03 ± 0.21 3.96 ± 0.17 1.81 ± 0.04 4.61 ± 0.11 2.69 ± 0.09 2.04 ± 0.04 24.03 ± 0.40 (n ¼ 12) 5.75 ± 0.14 (n ¼ 6)

0.29 n.d. 9.99 ± 0.17 3.98 ± 0.10 1.79 ± 0.03 4.49 ± 0.04 2.78 ± 0.07 2.10 ± 0.03 25.11 ± 0.28 (n ¼ 12) 5.69 ± 0.21 (n ¼ 6)

NB: HAC e acetic acid; HPr e propionic acid; HBu e butyric acid; HVa e valeric acid; n.d. e not detected, tVFA total VFA;, n e sample size. a Average VFA and pH measurement during fermentation steady state (day 1431).

Table 3b Statistical analysis of mean values of VFAs and pH between reactors. Comparative analysis

VFA

pH

t (df) NI vs N0 HI vs NI HI-CM vs HI NI-CM vs NI HI-CM vs NI-CM NI-RMN vs NI NI-RMN vs HI NI-RMN vs NI-CM NI-RMN vs HI-CM

6.28 7.36 3.98 8.81 1.34 13.43 4.97 2.22 1.11

(24) (24) (22) (24) (22) (24) (22) (22) (22)

p

Cohen’s d

t (df)

p**

Cohen’s d

0.000 0.000 0.001 0.000 0.196 0.000 0.000 0.037 0.277

2.429 2.884 1.624 3.424

7.22 2.23 1.73 1.21 3.16 1.51 4.27 0.37 3.45

0.000 0.050 0.114 0.255 0.010 0.162 0.002 0.721 0.006

4.169 1.287

5.299 2.029 0.906

(10) (10) (10) (10) (10) (10) (10) (10) (10)

1.826 2.464 1.990

NB: t e t-value, df e degrees of freedom, p e value. **Statistics carried out at 95% confidence interval.

the NI and N0 reactors, and between the HI and NI were all very large (Cohen’s d > 0.80) (Table 3b). The results indicated that the selection of inoculum affected the yield of VFAs. The trend of VFA production in NI and HI reactors, suggested the existence of similar microbial species or pattern of metabolism, contrary to that observed in the N0 reactor (Fig. 4). The start-up process of high-loading of sucrose, resulting in low pH in the NI reactor, was potentially capable of selecting spore-forming microflora population (possibly of Clostridrium or Bacillus species) comparable to that obtained in the thermal pre-treatment set-up (HI reactor). As highlighted by Hawkes et al. (2002) the heat-shock technique used in the pre-treatment of inoculum have the potential of destroying non-spore-forming facultative anaerobic microorganisms, in addition to the methanogens, reducing the system capacity to consume any traces of oxygen, and hence limiting biogas production by removal of an inhibitor, and also other anaerobic metabolisms. The selection of acidogenic bacteria using low pH resulting from high organic loading, leads to higher microbial diversity, due to lack of chemical or physical pre-treatment (Zhu and land, 2006). Although the HI gave a higher yield of VFAs with an Be increased quantity of longer chain acids than the NI, it can be suggested that the use of untreated inoculum with an initial sugar shock and additional micronutrients is a viable technique for the optimisation of VFA production and with a resultant increase ratio of acetic acid. 3.2. Enhancing VFA production with nutrients Table 4a shows that the amount of soluble COD in the digested sludge (used as inoculum) increased from 1.99 ± 0.05 g/l to 6.49 ± 0.28 g/l, representing more than 3 fold increase due to the thermal treatment. The thermal treatment also increased the bioavailability of carbohydrates, proteins (Table 4a) and trace minerals (Table 5). Table 4b shows the VFA equivalent of the total

COD in the CM and the RMN solutions. The CM was added at a concentration of 0.28 g/kgVS, indicating a concentration of 0.018 g/ l, representing 0.0018% (w/w). By comparing major trace minerals such as Ca and Na, it was observed that the concentration of trace elements in the RMN was generally 140 times less than the amount in the C. M stock. The quantity of RMN added was increased in an attempt to matchup the concentration in the CM added. The RMN was therefore added at a concentration of 39.2 g/kgVS (2.539 g/l, indicating a percentage concentration of 0.253% (w/w) of the substrate added). The amount of COD in the CM and RMN added was 0.0106 and 0.0164 g/l, equivalent to 0.0062 and 0.0096 g/l VFA, respectively. VFA production was expressed as COD equivalent (g COD) by converting VFA yield to g COD using the following conversion factors: 1.07 g COD g1 acetic acid, 1.51 g COD g1 propionic acid, 1.82 g COD g1 butyric acid and 2.04 g COD g1 valeric acid. The VFA equivalent of the COD added due to the CM and RMN additions were then subtracted from the total VFA produced in the CM and RMN reactors to obtain the VFA yielded through the actual fermentation. As shown in Table 4b, the quantity of COD in the CM stock was high, primarily because the trace minerals formulation was chelated in high concentrated liquid form. The COD added due to CM and RMN were however, minimal (Table 4b), having a VFA equivalent contribution of less than 0.4% of the total VFA yield. 3.2.1. The role of trace minerals in VFA production Micronutrients including Co, Ni, Mo and Fe have been reported to be important in the metabolic and enzymatic activities of methanogenic microbes (Zhang et al., 2012). The negative effect of the deficiency of the micronutrients in AD has been widely referred, but the exact concentrations needed significantly differ depending on the type of substrate and operating conditions. The concentration of Co, Ni, Mo and Fe in the CM added in the fermentation process in this study, were 0.075, 0.159, 0.082 and 0.296 mg/l

272

P.J. Kumi et al. / Water Research 100 (2016) 267e276

a

30

VFA yield (g VFACOD L-1)

25

20

15

10

5

0 0

5

10

15

20

25

30

35

Time (days) N0

NI

HI

Time 0: initial TH-WAS substrate feeding

b 30

25

VFA yield (g VFACOD

L-1)

20

NI 15

HI

NI-CM 10

HI-CM

5

0 0

5

10

15

20

25

30

35

Time (days) Time 0: initial TH-WAS substrate feeding

Fig. 3. a. Comparative yield of VFA e effect of inoculum pre-treatment with high organic loading. b. Comparative yield of VFA e effect of inoculum pre-treatment and micronutrients addition.

100%

n-Valeric Acid

90%

i-Valeric Acid

VFA production

80%

n-Butyric Acid

70%

i-Butyric Acid

60%

Propionic Acid

50%

AceƟc Acid

40%

30% 20%

10% 0% N0

NI

HI

HI-CM

NI-CM

NI-RMN

Fig. 4. Ratio of speciated volatile fatty acids production during steady state (day 14e31) of the fermentation of TH-WAS.

(Table 5). The concentrations of the same elements in the designated amount of the RMN added to the reactor were 0.071, 0.106, 0.071, and 1.548 mg/l, respectively, compared to 0.202, 0.049, 0.040 and 0.201 mg/l, respectively, in the widely reported Wolin et al. (1963) recipe. The concentration of Co was relatively lower, whereas Ni, Mo and Fe in both CM and RMN were above the formulations of Wolin et al. (1963).

Table 6 shows the quantity of trace minerals available in the NI, NI-CM and NI-RMN reactors after substrate feeding. The concentration of Co, Ni, Mo and Fe in the NI reactor were, 0.043, 0.019, 0.006 and 15 mg/l, respectively. The concentration of the same elements in the NI-RMN reactor was 0.076, 0.149, 0.026 and 26.06 mg/l, compared concentrations of 0.081, 0.279, 0.035 and 21.85 mg/l, of the same elements respectively, in the NI-CM reactor. As shown in Table 6 the concentration of all the minerals (except Al, Sb and Se) increased significantly after supplementation (comparing NI to NI-CM and NI-RMN). The concentration of Co, Ni, and Mo in all the reactors were within the stimulatory concentrations, and outside the inhibitory concentrations for anaerobic digestion, as highlighted by RomeroGüiza et al. (2016). The concentrations of Fe in the all the reactors were particularly high, when compared to the stimulatory concentration of <0.3 mentioned by Romero-Güiza et al. (2016). Fe has been widely used by researchers not only for its capacity as a cofactor for enzymatic reactions in anaerobic digestion but also for its ability to reduce oxidativeereductive potential in anaerobic digestion (Romero-Güiza et al., 2016). The inhibitory concentrations of Fe on anaerobic hydrolytic-acidogenic fermentation have however not been clearly defined by current research. High concentrations of Fe (III) oxide (16 g L1) was reported by Coates et al. (2005) to have a negative effect on VFA production. The enhanced yield of VFA in the supplement reactors showed that the Fe concentration at 22e26 mg/l in this study, gave an indication that the concentrations used did not have inhibitory effect on VFA production. The evaluation of the concentrations of the micronutrients in the reactors (NI, NI-RMN and NI-CM) after substrate feeding and in effluents further emphasized that the micronutrients addition have likely contributed to the differences in the yield of VFAs in the reactors (Table 6). The concentrations of the micronutrients in the effluent were generally lower than the concentrations in the reactors after substrate feeding. The trend in micronutrients uptake was, however, not clearly defined in the reactors. As shown in Table 6, no reduction or increase was observed in the elements including Al, Cd, Sb and Se. The lack of difference between the concentration of these elements at substrate feeding and in the effluent of the reactors could be attributed to the low concentrations of these elements, i.e. close to the lower range of the detection limit of the ICP-OES analytical technique used. It can also be said that the contributions of these elements in the difference in the VFA yields in the reactors were not significant. The elements with a notable concentration reduction in the reactors, and therefore potentially contributed more in the difference in the VFA yields included Co, Cu, Fe, Ni, Ti, S and Zn. As emphasized by Xie et al. (2015), the degree of uptake of Zn, Cu, Ni and Cd in an anaerobic reactor is highly dependent on the pH of the reactor and not necessary the degree of importance. These elements could however be important indicators in the performance of acidogenic fermentation because they are also pH dependant. Smith and Carliell-Marquet (2008) also emphasized the importance of Fe in AD, after they observed a positive linear relationship between total Fe concentration and biogas yield. Florencio et al. (1993) also highlighted the importance of Co in acidogenic and methanogenic conversions. They observed that the stimulation of methanogenesis due to the addition of Co alone was comparable to the stimulation when a complete trace elements mixture was added. Lin (1993) defined the importance of the relative toxicity of heavy metals to VFA production as: Cu > Zn > Cd > Ni. The concentration of Cu in the NI, NI-RMN and NI-CM substrates were 0.035, 0.051 and 0.062 mg/l, respectively; whilst its concentration in the effluents of the same reactors were 0.013, 0.12, 0.01 mg/l,

P.J. Kumi et al. / Water Research 100 (2016) 267e276

273

Table 4a Release of carbohydrates and proteins after thermal treatment of sludge compared to the concentrations in the commercial micronutrient stock.

Soluble carbohydrate (mg/l) soluble Protein (mg/l) Soluble COD (g/l) a b

Digested sludgea

Recovered microbial nutrients (RMN)b

Commercial micronutrients (CM)

0.047 ± 0.0004 0.146 ± 0.002 1.99 ± 0.05

0.797 ± 0.0003 3.350 ± 0.223 6.49 ± 0.28

28.54 ± 0.65 37.05 ± 1.08 590.95 ± 11.20

Liquor of untreated digested sludge used as inoculum. Thermally treated digested sludge.

Table 4b The profile of COD in the commercial micronutrients stock and the recovered microbial nutrients stock.

COD (g/l) Amount added (%) COD added (g/l) VFA equivalent of COD added (g/l)

Recovered microbial nutrients (RMN)

Commercial micronutrients (CM)

6.49 ± 0.28 0.2530 0.0164 0.0096

590.95 ± 11.20 0.0018 0.0106 0.0062

COD concentration was converted to total VFA by a conversion factor of 1.72g COD/g total VFA.

Table 5 The profile of trace elements in the micronutrients solution compared with Wolin et al. (1963).

Al Ca Cd Co Cu Fe K Mg Mn Mo Na Ni P S Sb Se Si Ti V Zn

RMN (mg/l)

CM (mg/l)

Wolin et al. (1963)

0.198 5.455 0.035 0.071 0.072 1.548 26.494 0.691 0.142 0.071 47.357 0.106 7.339 29.211 0.354 0.319 7.853 0.354 0.106 0.177

<0.001 0.002 <0.001 0.076 0.121 0.296 0.183 0.016 0.131 0.082 0.331 0.159 <0.001 0.003 <0.001 0.034 0.001 <0.001 <0.001 0.176

0.010 0.361 na 0.202 0.025 0.201 0.015 2.958 1.625 0.040 3.934 0.049 na 6.161 na 0.005 na na na 0.227

NB: na not applicable.

respectively. The results therefore indicated a Cu reduction of approximately 63, 76 and 84% in the NI, NI-RMN and NI-CM reactors respectively. Although significant uptake was observed, the quantity was very low, and therefore it was not expected to have a toxic effect on the reactor. Lin (1993) emphasized that Ni concentration of less than 300 mg/l was important for a healthy acidogenic environment. The concentrations of Ni, in all the reactors (NI, NIRMN and NI-CM), as shown in Table 6, were far lower than the toxicity limit defined by Lin (1993). The uptake of Zn was highest in the fermentation process (more than 80% uptake in all the reactors). The concentration was however low and therefore not expected to have any significant toxicity effect on the fermentation process. It is a common trend that the role of micronutrients in AD is focused entirely on certain key elements such as Co, Fe and Ni, but it is important to note that the right balance of all the micronutrients in an anaerobic fermentation/digestion process may as well ensure a healthy metabolic environment. As highlighted by Smith and Carliell-Marquet (2008) the overdose of Fe or Al in the chemical P removal, not only caused toxicity in the AD process, but also the limitation of the bioavailability of P. Thus, the bioavailability of one

Table 6 Profile of micro-minerals in the reactor after substrate feeding and in effluent of the reactors.

Al Ca Cd Co Cu Fe K Mg Mn Mo Na Ni P S Sb Se Si Ti V Zn

Concentration of minerals in reactor after feeding (mg/l)

Effluent (mg/l)

NI

NI-RMN

NI-CM

NI

NI-RMN

NI-CM

<0.056 54.8 <0.013 0.043 0.035 15 104 67.28 0.022 0.006 79.6 0.019 113.9 152 <0.06 <0.06 15.3 0.05 0.004 0.015

<0.56 135.4 <0.013 0.076 0.051 26.05 292 82.98 0.902 0.026 185.1 0.149 185.9 171.44 <0.06 <0.06 37.7 0.057 0.034 0.054

<0.56 139.7 <0.013 0.081 0.062 21.85 283 81.68 1.112 0.035 181.6 0.279 186.9 162.44 <0.06 <0.06 34.6 0.053 0.034 0.245

<0.056 82.5 <0.013 0.041 0.013 13.4 105 69.2 0.02 <0.006 82.2 0.01 141 110 <0.06 <0.06 18.6 0.031 0.003 0.003

<0.56 91.5 <0.013 0.042 0.012 14.1 211 94.6 1.07 <0.02 84.5 0.1 228 110 <0.06 <0.06 18.9 0.03 0.03 0.006

<0.56 81.4 <0.013 0.054 0.01 12 190 89.1 1.03 <0.02 76.7 0.22 208 102 <0.06 <0.06 17 0.027 0.04 0.022

NB e Concentration of minerals taken before feeding (effluents) and after feeding on day 28.

element is dependent on the concentration of another. This study is set on the basis that the micronutrients extracted from the microbial culture used in a near optimal fermentation/ digestion of a particular substrate may provide the most suitable balance of the micronutrients for the fermentation/digestion of the same substrate.

3.2.2. The effect of micronutrients addition on the VFA production As shown in Fig. 3b, the average total VFA produced in the NI-CM reactor at steady-state fermentation was 24.03 ± 0.42 gVFACOD L1, approximately 21% higher than the yield obtained from the NI reactor (19.84 ± 0.28 gVFACOD L1). The results of an independent ttest (Table 3b) showed that the mean difference between VFA yields from NI-CM and NI reactors were statistically significant, t ¼ 8.810, p ¼ 0.000. The effect size, Cohen’s d (3.696) > 0.80 were large, indicating that the distribution of the mean values of VFA yields between the two reactors were very different from one another. The average total VFA yield from the HI-CM reactor, at steady-

P.J. Kumi et al. / Water Research 100 (2016) 267e276

state fermentation was 24.70 ± 0.28 gVFACOD L1, compared to the average total VFA yield of 22.98 ± 0.35 gVFACOD L1 observed in the HI reactor, indicating a yield increase of approximately 7% (Table 3a). The results of an independent t-test (Table 3b) showed that the mean difference between VFA yields from HI-CM and HI reactors were statistically significant, t ¼ 3.98, p ¼ 0.001. The effect size, Cohen’s d (1.624) > 0.80 were large, indicating that the distribution of the mean values of VFA yields between the two reactors was significantly different. The VFA yield difference observed due to the addition of micronutrients (comparing VFA yields from NI and NI-CM reactors) was found to be higher than the VFA yield difference observed in the same experimental setup, but with the thermal pre-treatment of the inoculum (comparing VFA yields from HI and HI-CM reactors). The addition of micronutrients was therefore found to be less effective when the inoculum was heat-treated. A comparison of the average total VFA produced in the HI-CM reactor (24.70 ± 0.28 gVFACOD L1) and that produced in the NICM reactor (24.03 ± 0.42 gVFACOD L1), showed a negligible percentage yield rise of 3% (yield increase within the error margin). The results of an independent t-test (Table 3b) showed that the mean difference between VFA yields from HI-CM and NI-CM reactors were not statistically significant, t ¼ 1.34, p ¼ 0.196. It can be considered from the results that the thermal treatment of the inoculum (HI reactor) caused lyses of the microbial cell walls resulting in the release of cytoplasmic inorganic ions and organic compounds, and the dissolution of the membrane bound extracellular polymeric substances (EPS), which supported the production of VFA. The yield increase between the NI and HI reactors was comparable to the yield increase between the NI and NI-CM reactors. The micronutrients and organic compounds released due to the dissolution of EPS and release of cytoplasmic inorganic ions were therefore found to cause a significant increase in VFA production, eclipsing the importance of the addition of micronutrients, when the inoculum was not thermally pre-treated. The most prominent VFA product in the effluent of N0, NI and HI, HI-CM, NI-CM and NI-RMN reactors was acetic acid, representing approximately 32, 35 and 37, 39, 37 and 40% of the total VFA respectively (Fig. 3a), consistent with the findings from MorganSagastume et al. (2011), who also observe an acetic acid fraction of 35e40%, when fermenting TH-WAS sludge at 42
C. The trend of VFA production in the N0 reactor, as illustrated in Fig. 4, was quite different from all the other reactors (NI, HI, HI-CM, NI-CM, NI-RMN). The major variation was observed in the trend of production of the acetic and n-valeric acids. Higher chain length VFAs production seemed to be more related to a lack of micronutrients availability. The trend of production of VFA correlated particularly well with the different environmental conditions, which likely impacted on the microbial populations in the anaerobic reactors. Williams et al. (2013), using the quantitative polymerase chain reaction technique to monitor the microbial populations in a full scale anaerobic digester treating food wastes, observed a significantly positive effect from an addition of micronutrients on the populations of bacteria and hydrogenotrophic methanogens. The negligible yield of H2 in the NI reactors could be linked to H2 consumption by the hydrogenotrophic methanogens or homoacetogenic microorganisms, which were potentially stimulated by the addition of micronutrients, according to the observation of Williams et al. (2013). Luo et al. (2011), however, observed that whereas hydrogenotrophic methanogens are limited at pH below 6 (as observed in NI, NI-CM and NI-RMN), homoacetogens were not affected. It can therefore be suggested that the micronutrients addition potentially aided the growth of acidogenic syntrophic bacteria (which facilitated propionic acid conversions to acetic acid) and homoacetogenic microorganisms (which facilitated

conversion of H2 and CO2 to acetate), resulting in improved VFA yield in the supplemented reactors (NI-CM and NI-RMN). The higher pH (6.75 ± 0.06) and partial methanogenesis observed in the N0 reactors gave indication of a possible presence of hydrogenotrophic methanogens that consumed H2 leading to a reduced H2 yield and consequently a lower VFA yield. The trend of production of VFA products in the NI-RMN reactor was closer to the VFA production trend in the NI-CM and HI-CM reactors, compared to the trend observed in the NI and HI reactors, indicating that the micronutrients availability have potentially allowed a conversion of higher chain VFAs to acetic acid, supported by a possible increase in growth/activity of homoacetogens. The selection of the acidogenic population via the high loading-low pH technique, used in this study, resulted in a relatively low pH and therefore contributed to a shift in the microbial metabolism towards the production of lower chain VFAs (particularly acetic acid) rather than hydrogen. 3.3. Recovered soluble microbial nutrients as a micronutrient source to enhance VFA production As shown in Fig. 5, the NI-RMN reactor resulted in an average VFA yield of 25.11 ± 0.28 gVFACOD L1. The yield was found to be approximately 27% higher than the yield obtained in the NI (control) reactor, with no RMN addition. The results of an independent t-test (Table 3b) showed that the mean difference between VFA yields from NI-RMN and NI reactors were statistically significant, t ¼ 13.43, p ¼ 0.000. The effect size, Cohen’s d (5.299) > 0.80 were large, indicating that the distribution of the mean values of VFA yields between the two reactors was significantly different. The VFA yield obtained from the NI-RMN reactor was comparable to the yield obtained from the NI-CM reactor (24.03 ± 0.42 gVFACOD L1). The yield difference of approximately 4% in favour of NI-RMN compared to the NI-CM, can be considered to be negligible (within the error margin). The results of an independent t-test (Table 3b), however, showed that the mean value difference between VFA yields from NI-RMN and NI-CM reactors were statistically significant, t ¼ 2.22, p ¼ 0.037. The effect size, Cohen’s d (0.906) > 0.80 were large, indicating that the distribution of the mean values of VFA yields between the two reactors was significantly different. The VFA yield obtained from the NI-RMN reactor was also found to be approximately 9% higher than the yield obtained from the HI

30

25

VFA yield (g VFACOD L-1)

274

20

15 NI HI

10

NI-CM NI-RMN

5

0

0

5

10

15

20

25

30

35

Time (days) 0: initial TH-WAS substrate feeding

Fig. 5. Comparative yield of VFA e effect of inoculum pre-treatment, commercial micronutrients (CM) addition and recovered microbial nutrients (RMN) addition.

P.J. Kumi et al. / Water Research 100 (2016) 267e276

reactor (22.98 ± 0.35 gVFACOD L1). The results of an independent ttest (Table 3b) showed that the mean difference between VFA yields from NI-RMN and HI reactors were statistically significant, t ¼ 4.97, p ¼ 0.000. The effect size, Cohen’s d (2.029) > 0.80 were large, indicating that the distribution of the mean values of VFA yields between the two reactors was significantly different. The EPS matrix, a complex high-molecular weight mixture of polymers surrounding the living cells, usually contains highmolecular weight secretions from the microorganism, inorganic ions, macromolecules, lipids, proteins, carbohydrates, nucleic acid and uronic acids (Sheng et al., 2010). The thermal treatment of the digested sludge used in the preparation of RMN may have resulted in the dissolution of the EPS and release of SMP, indicated by the increased concentration of soluble COD, carbohydrates, proteins (Table 4a) and bioavailability of trace minerals (Table 5) in the RMN compared to the control (digested sludge centrate). The enhanced VFA production in the NI-RMN, compared to NI could therefore be attributed to the bioavailability of inorganic ions (trace minerals) and organic macro-micronutrients (including: proteins and carbohydrates) from the RMN addition. The quantity of VFA produced from the TH-WAS with no micronutrients addition was 16.56 ± 0.46 gVFACOD L1, which was found to be consistent with the quantity observed by MorganSagastume et al. (2011) (14.2 ± 0.6 gVFACOD L1). MorganSagastume et al. observed a 15e25% VFA yield increase when the substrate was changed from TH-WAS to a thermally hydrolysed sludge composed of a mixture of primary sludge and WAS. The primary sludge is a much easier to hydrolyse substrate than WAS and provides more easily available nutrients and an organic carbon source which would likely also reduced pH more effectively, resulting in the improved VFA yield. As emphasized by Ucisik and Henze (2008), primary sludge (because of availability of excess nutrients) is easier to digest compared to WAS. The importance of the availability of additional nutrients was confirmed in this study, as observed in the NI-CM and NI-RMN, resulting in a VFA yield of 24e25 gVFACOD L1 (taking into consideration the two runs) i.e. 21e27% higher than the yield obtained in the no micronutrients control (NI). The use of RMN as micronutrients source, instead of CM, may give relatively higher VFA yield compared to the primary-secondary sludge mixture technique employed by Morgan-Sagastume et al. (2011) in plants whereby sludges are of secondary sewage treatment nature only, such as in sequencing batch reactors. Taking into consideration the additional VFA yield generated by the mixture in MorganSagastume et al. study in comparison with TH-WAS only, the use of RMN could further improve the yield of VFAs obtained even from primary and secondary sewage sludge mixtures, as substrate. The use of RMN is expected to also help reduce the overall operational cost of plants by facilitating an improved acidification and even a digestion process without the need to add more micronutrients, resulting in a further help to reduce the environmental burden of sludges when disposed to land due to their lower metal content. The use of RMN as micronutrients will also improve the overall economic benefits of AD by obtaining value added products (chemicals) from the digested sludge, perfecting the story of AD in terms of life cycle assessment and making RMN a potential media for the growth of numerous microbial species within bio refinery platforms. RMN as a micronutrients source can also be used in secondary processes such as methane production, PHA production and in BES. 4. Conclusion  The selection of acidogenic inoculum by reducing pH using a high dose of sucrose was found to be effective for VFA

275

production from the fermentation of thermally hydrolysed WAS, increasing VFA yield by approximately 20%, compared to the control.  Thermal treatment of inoculum can enhance VFA production, approximately 16% higher than that obtained from low pHsucrose inoculum pre-treatment. The amount of VFA produced using thermally pre-treated inoculum was comparable (3% difference) to the quantity produced when commercial micronutrients without thermal pre-treatment of inoculum was used.  The addition of recovered microbial nutrients, as nutrient source in the fermentation associated with sucrose shock inoculum selection resulted in a VFA yield of 25.11 ± 0.27 g VFACOD L1, which was comparable (4% difference) to the yield obtained from the use of commercial micronutrients; approximately 27% higher than the control, and about 9% higher than the VFA yield from thermal inoculum pretreatment.  The study emphasizes the importance of exploiting the benefits of anaerobically digested sludge as microbial activity enhancing micronutrient sources. Acknowledgements The authors would like to acknowledge the financial support provided by the European Regional Development Fund (ERDF) (HE14161001 SUPER CIRP, HE14151009 KTC AAPBS) through the A4B, AAPBS KTC, the SuPERPHA Collaborative Industrial Research Project (CIRP) and all associated partners. References Coates, J.D., Cole, K.A., Michaelidou, U., Patrick, J., Mcinerney, M.J., Achenbach, L.A., 2005. Biological control of hog waste odor through stimulated microbial Fe (III) reduction. Appl. Environ. Microbiol. 71, 4728e4735. http://dx.doi.org/10.1128/ AEM.71.8.4728. Cruwys, J.A., Dinsdale, R.M., Hawkes, F.R., Hawkes, D.L., 2002. Development of a static headspace gas chromatographic procedure for the routine analysis of volatile fatty acids in wastewaters. J. Chromatogr. A 945 (1e2), 195e209. http:// dx.doi.org/10.1016/S0021-9673(01)01514-X. Dubois, M., Gilles, K., Hamilton, J., Rebers, P., Smith, F., 1956. Colorimetric method for determination of sugars and related substances. Anal. Chem. 28 (3), 350e356. http://dx.doi.org/10.1021/ac60111a017. Fassel, V.A., Knlseley, R.N., 1974. Inductively coupled plasma optical emission spectroscopy. Anal. Chem. 46 (13) http://dx.doi.org/10.1021/ac60349a023. Feeherry, F.E., Munsey, D.T., Rowley, D.B., 1986. Thermal inactivation and injury of Bacillus stearothermophilus spores. Appl. Environ. Microbiol. 53 (2), 365e370. http://aem.asm.org/content/53/2/365.full.pdf. Feng, X.-M., Karlsson, A., Svensson, B.H., Bertilsson, S., 2010. Impact of trace element addition on biogas production from food industrial waste -linking process to microbial communities. FEMS Microbiol. Ecol. 74, 226e240. http://dx.doi.org/ 10.1111/j.1574-6941.2010.00932.x. Florencio, L., Jeni cek, P., Field, J.A., Lettinga, G., 1993. Effect of cobalt on the anaerobic degradation of methanol. J. Ferment. Bioeng. 75 (5), 368e374. http:// dx.doi.org/10.1016/0922-338X(93)90136-V. Guwy, A.J., Dinsdale, R.M., Kim, J.R., Massanet-Nicolau, J., Premier, G., 2011. Fermentative biohydrogen production systems integration. Bioresour. Technol. 102, 8534e8542. http://dx.doi.org/10.1016/j.biortech.2011.04.051. Hawkes, F.R., Dinsdale, R., Hawkes, D.L., Hussy, I., 2002. Sustainable fermentative hydrogen production: challenges for process optimisation. International Journal of Hydrogen Energy. Biohydrogen 27, 1339e1347. http://dx.doi.org/10.1016/ S0360-3199(02)00090-3. Kedia, G., Passanha, P., Dinsdale, R.M., Guwy, A.J., Esteves, S.R., 2014. Evaluation of feeding regimes to enhance PHA production using acetic and butyric acids by a pure culture of Cupriavidus necator. Biotechnol. Bioprocess Eng. 19 (6), 989e995. http://dx.doi.org/10.1007/s12257-014-0144-z. Kyazze, G., Martinez-Perez, N., Dinsdale, R., Premier, G.C., Hawkes, F.R., Guwy, A.J., Hawkes, D.L., 2006. Influence of substrate concentration on the stability and yield of continuous biohydrogen production. Biotechnol. Bioeng. 93 (5), 971e979. http://dx.doi.org/10.1002/bit.20802. Lee, W.S., Chua, A.S.M., Yeoh, H.K., Ngoh, G.C., 2014. A review of the production and applications of waste-derived volatile fatty acids. Chem. Eng. J. 235, 83e99. http://dx.doi.org/10.1016/j.cej.2013.09.002. Lin, C., 1993. Effect of heavy metals on acidogenesis in anaerobic digestion. Water Res. 27 (1), 147e152. http://dx.doi.org/10.1016/0043-1354(93)90205-V. Luo, G., Karakashev, D., Xie, L., Zhou, Q., Angelidaki, I., 2011. Long-term effect of inoculum pretreatment on fermentative hydrogen production by repeated batch cultivations: homoacetogenesis and methanogenesis as competitors to

276

P.J. Kumi et al. / Water Research 100 (2016) 267e276

hydrogen production. Biotechnol. Bioeng. 108 (8), 1816e1827. http://dx.doi.org/ 10.1002/bit.23122. Maharaj, I., Elefsiniotis, P., 2001. The role of HRT and low temperature on the acidphase anaerobic digestion of municipal and industrial wastewaters. Bioresour. Technol. 76, 191e197. http://dx.doi.org/10.1016/S0960-8524(00)001280. Massanet-Nicolau, J., Dinsdale, R., Guwy, A., 2008. Hydrogen production from sewage sludge using mixed microflora inoculum: effect of pH and enzymatic pre-treatment. Bioresour. Technol. 99, 6325e6331. http://dx.doi.org/10.1016/ j.biortech.2007.12.012. Massanet-Nicolau, J., Dinsdale, R., Guwy, A., Shipley, G., 2013. Use of real time gas production data for more accurate comparison of continuous single-stage and two-stage fermentation. Bioresour. Technol. 129, 561e567. http://dx.doi.org/ 10.1016/j.biortech.2012.11.102. Massanet-Nicolau, J., Guwy, A., Dinsdale, R., Premier, G., Esteves, S., 2009. Production of hydrogen from sewage biosolids in a continuously fed bioreactor: effect of hydraulic retention time and sparging. Int. J. Hydrogen Energy 35, 469e478. http://dx.doi.org/10.1016/j.ijhydene.2009.10.076. Morgan-Sagastume, F., Pratt, S., Karlsson, A., Cirne, D., Lant, P., Werker, A., 2011. Production of volatile fatty acids by fermentation of waste activated sludge pretreated in full-scale thermal hydrolysis plants. Bioresour. Technol. 102 (3), 3089e3097. http://dx.doi.org/10.1016/j.biortech.2010.10.054. Romero-Güiza, M.S., Vila, J., Mata-Alvarez, J., Chimenos, J.M., Astals, S., 2016. The role of additives on anaerobic digestion: a review. Renew. Sustain. Energy Rev. 58, 1486e1499. http://dx.doi.org/10.1016/j.rser.2015.12.094. Shana, A., Asaadi, M., Morchoisne, T., Marsh, P., 2003. Acid phase digestion pretreatment and its impact on conventional mesophilic anaerobic digestion. In: 8th European Biosolids and Organic Residuals Conference, Wakefield, November 2003. Sheng, G., Yu, H., Li, X., 2010. Extracellular polymeric substances (EPS) of microbial aggregates in biological wastewater treatment systems: a review. Biotechnol. Adv. 28, 882e894. http://dx.doi.org/10.1016/j.biotechadv.2010.08.001.

Smith, J.A., Carliell-Marquet, C.M., 2008. The digestibility of iron-dosed activated sludge. Bioresour. Technol. 99, 8585e8592. http://dx.doi.org/10.1016/ j.biortech.2008.04.005. Ucisik, A.S., Henze, M., 2008. Biological hydrolysis and acidification of sludge under anaerobic conditions: the effect of sludge type and origin on the production and composition of volatile fatty acids. Wat. Res. 42, 3729e3738. http://dx.doi.org/ 10.1016/j.watres.2008.06.010. Wang, C.C., Chang, C.W., Chu, C.P., Lee, D.J., Chang, B.-V., Liao, C.S., 2003. Producing hydrogen from wastewater sludge by Clostridium bifermentans. J. Biotechnol. 102, 83e92. http://dx.doi.org/10.1016/S0168-1656(03)00007-5. Williams, J., Williams, H., Dinsdale, R., Guwy, A., Esteves, S., 2013. Monitoring methanogenic population dynamics in a full-scale anaerobic digester to facilitate operational management. Bioresour. Technol. 140, 234e242. http:// dx.doi.org/10.1016/j.biortech.2013.04.089. Wolin, E.A., Wolin, M.J., Wolfe, R.S., 1963. Formation of methane by bacterial extracts. J. Biol. Chem. 238, 2882e2886. http://www.jbc.org/content/238/8/2882. short. Xie, S., Ma, Y., Strong, P.J., Clarke, W.P., 2015. Fluctuation of dissolved heavy metal concentrations in the leachate from anaerobic digestion of municipal solid waste in commercial scale landfill bioreactors: the effect of pH and associated mechanisms. J. Hazard. Mater. 299, 577e583. http://dx.doi.org/10.1016/ j.jhazmat.2015.07.065. Yuan, Q., Sparling, R., Oleszkiewicz, J.A., 2009. Waste activated sludge fermentation: effect of solids retention time and biomass concentration. Water Res. 43, 5180e5186. http://dx.doi.org/10.1016/j.watres.2009.08.019. Zhang, L., Ouyang, W., Lia, A., 2012. Essential role of micronutrients in continuous anaerobic digestion of food waste. In: Procedia Environmental Sciences, the Seventh International Conference on Waste Management and Technology (ICWMT 7), vol. 6, pp. 102e111. http://dx.doi.org/10.1016/j.proenv.2012.10.014. land, M., 2006. Evaluation of alternative methods of preparing hydrogen Zhu, H., Be producing seeds from digested wastewater sludge. Int. J. Hydrogen Energy 31, 1980e1988. http://dx.doi.org/10.1016/j.ijhydene.2006.01.019.