- Email: [email protected]
Impact of dewatering technologies on specific methanogenic activity
w a t e r r e s e a r c h 8 2 ( 2 0 1 5 ) 7 8 e8 5
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/watres
Impact of dewatering technologies on specific methanogenic activity Damien J. Batstone*, Yang Lu, Paul D. Jensen Advanced Water Management Centre (AWMC), The University of Queensland, St Lucia, 4072, Brisbane, Queensland, Australia
article info
abstract
Article history:
Dewatering methods for recuperative thickening and final dewatering can potentially
Received 30 November 2014
impact methanogenic activity and microbial community. This influences both the feasi-
Received in revised form
bility of recuperative thickening to increase solids residence time within a digester, and the
20 March 2015
utilisation of dewatered digestate as inoculum for new digesters. Thickening technology
Accepted 2 April 2015
can reduce methanogenic activity through either air contact (rotary drum, DAF, or belt filter press), or by lysing cells through shear (centrifuge). To assess this, two plants with recuperative thickening (rotary drum) in their anaerobic digester, and five without recu-
Keywords:
perative thickening, had specific methanogenic activity tested in all related streams,
Recuperative
including dewatering feed, thickened return, final cake, and centrate. All plants had high
Thickening
speed centrifuges for final dewatering. The digester microbial community was also
Shear
assessed through 16s pyrotag sequencing and subsequent principal component analysis
Specific methanogenic activity
(PCA). The specific methanogenic activity of all samples was in the expected range of 0.2
Pyrotag sequencing
e0.4 gCOD gVS1d1. Plants with recuperative thickening did not have lower digester activity. Centrifuge based dewatering had a significant and variable impact on methanogenic activity in all samples, ranging between 20% and 90% decrease but averaging 54%. Rotary drum based recuperative thickening had a far smaller impact on activity, with a 0% perpass drop in activity in one plant, and a 20% drop in another. However, the presence of recuperative thickening was a major predictor of overall microbial community (PC1, p ¼ 0.0024). Microbial community PC3 (mainly driven by a shift in methanogens) was a strong predictor for sensitivity in activity to shear (p ¼ 0.0005, p ¼ 0.00001 without outlier). The one outlier was related to a plant producing the wettest cake (17% solids). This indicates that high solids is a potential driver of sensitivity to shear, but that a resilient microbial community can also bestow resilience. Sensitivity of methanogens to centrifuging does not rule out centrifuges for recuperative thickening (particularly where hydrolysis is rate-limiting), but may impose a maximum return rate to avoid digester failure. © 2015 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: þ61 3346 9051; fax: þ61 7 3365 4726. E-mail address: [email protected] (D.J. Batstone). http://dx.doi.org/10.1016/j.watres.2015.04.005 0043-1354/© 2015 Elsevier Ltd. All rights reserved.
w a t e r r e s e a r c h 8 2 ( 2 0 1 5 ) 7 8 e8 5
1.
Introduction
Recuperative thickening is a process for intensifying anaerobic digesters, by removing water from digestate, and returning the high-solids concentrate back to the digester. It thereby increases effective solids retention time over hydraulic retention time (Reynolds et al., 2001; Vanyushina et al., 2012). It is particularly suited to primary sludge, which has a relatively high degradability extent, but which is also relatively slow to degrade, and generally poorly suited material with a limited degradability extent, since this will result in the recycling of large amounts of inert material (see model provided supplementary information SI). Recuperative thickening is a relatively low cost technique for intensification, with well understood fundamentals, but there is very references found in the formal literature. A key practical concern is the possibility of damaging anaerobic microbes, particularly methanogenic archaea during the thickening stage. The impact of recuperative thickening on methanogenic activity is of critical importance, since the digester contents may be passed through the thickening equipment multiple times before it leaves the digester for final dewatering. The other major process of water removal from sludge is final dewatering, and indeed, centrifuges can be used for both recuperative thickening and final dewatering. The impact of final dewatering on activity is also important since firstly, dewatered sludge may be used as inoculum in other digesters, and secondly, the activity of dewatered material potentially determines methane production during storage. Both recuperative thickening and final dewatering can be regarded as comparable processes. In the case of recuperative thickening, digester sludge is thickened to 6%e12% through a range of techniques, including centrifuges and belt filter presses (Tchobanoglous et al., 2003), dissolved air flotation (Vanyushina et al., 2012), gravity belt thickeners (Reynolds et al., 2001), and rotary drum thickeners (Tchobanoglous et al., 2003). These can be generally classed by the DS achieved and the potential impact on the biology as shown in Table 1. Final dewatering depends on high-g and standard centrifuges, or belt filter presses to produce a final cake at >12% DS (Albertson et al., 1987) whereas recuperative thickening technologies target lower DS. The major factors that may influence activity are liquid shear and oxygen exposure (Table 1). Flocculant is not inhibitory at normal doses, and indeed, may partially degrade
Table 1 e Dewatering and thickening equipment summary. Equipment High-g centrifuge Centrifuge Belt filter press Rotary drum Baleen filter Dissolved air flotation Gravity belt thickening
Performance (% DS)
Biochemical impact
18%e30% 15%e25% 12%e25% 8%e15% 5%e8% 5%e8% 4%e6%
High shear Shear Oxygen, mild shear Oxygen Oxygen, mild shear High oxygen Oxygen
From Albertson et al. (1987), Tchobanoglous et al. (2003).
79
(Chang et al., 2001). Liquid shear occurring at impeller tip velocities >2 m s1 (shear rate e G~2000 s1 in a 1 mm tip field) are of concern since this lyse susceptible organisms, and may disrupt microbial syntrophic associations (Deveci, 2002; Ghyoot and Verstraete, 1997), while oxygen is toxic to methanogenic archaea in particular (Madigan et al., 2009). High-g equipment such as centrifuges (G 5000 s1 (Muller, 2006))are more likely to expose organisms to shear, while dissolved air filtration, and rotary drum thickeners in particular are more likely to expose organisms to oxygen (Table 1). The issue of oxygen exposure during thickening for recuperative purposes has been previously considered. Conklin et al. (2007) (Conklin et al., 2007) assessed the impact of oxygen exposure (via sparging) as a proxy for dissolved air flotation and actual samples from a gravity belt thickener (GBT) to assess the potential use of either for recuperative thickening. They found (via activity) that mild exposure (i.e., GBT) had no impact on activity while high exposure (i.e., sparging) only had an impact, but would only be of effect at high relative recirculation rates, or where the reactor was overloaded. The impact of centrifuge dewatering on activity has not been previously considered, either for recuperative thickening, or for final dewatering, and particularly, the impact of shear on methanogens has not been investigated. However, Deveci (2002) found that shearing had a strong impact on viability of aerobic acidophilic organisms particularly at solids concentrations above 10%. The finding was that shear alone was less likely to result in loss of activity, but that at high solids, poor momentum dissipation, and hence a high velocity gradient (i.e., shear) would result in loss of activity. Application of membrane based methods to selectively remove permeate have consistently found reduced activities at higher recirculation rates (Ghyoot and Verstraete, 1997), and high shear in a membrane configuration is correlated with a loss of acetoclastic methanogens (Padmasiri et al., 2007). As seen in Table 1, particularly mechanical dewatering equipment operates in an environment of high-shear and high solids, and the impact of dewatering equipment on anaerobes and particularly methanogens is unknown. While there is relevant incidental literature, there is very little available on the impact of dewatering on anaerobic systems, and particularly on activity, more so for the purpose of recuperative thickening. This is surprising given the relatively feasibility and simple application of recuperative thickening to intensify anaerobic processes. This paper addresses the key gap of impact of shear (centrifuge) based dewatering and air based (rotary drum) thickening on methanogenic activity, and also assesses the impact of long-term operation using rotary drum thickening on microbial community.
2.
Methods
2.1.
Digesters and sampling
27 samples were taken from 7 sites in the Sydney greater metropolitan area (plant names given in Table 2). These included two sites with recuperative thickening (A, B e six samples each), and five sites without recuperative thickening
80
w a t e r r e s e a r c h 8 2 ( 2 0 1 5 ) 7 8 e8 5
Table 2 e Treatment processes in the various plants studied. Ident A B C D E F G a b
Site North head Bondi Cronulla Malabar Glenfield Liverpool Wollongong
Feed to digester Primary Primary Blended Primary Blended Blended Blended
Loading kgVS/m3 d1
Digester HRT (d)
Recup
3.0 2.3 3.4 2.6 1.8 1.8 2.7
13 10 10 12 17 19 13.5
RDTb RDT N/A N/A N/A N/A N/A
(70% primary)a (70% primary) (70% primary) (60% primary)
Indicates load by mass dry solids. All plants use high-speed centrifuges for final dewatering; RDT ¼ rotary drum thickener.
(C, D, E, F, G e 3 samples each). Sites A and B used rotary drum recuperative thickening, and high-g centrifuges for final dewatering, while sites C-G used high-g centrifuges for final dewatering. Solids analysis results were compared with historical data to ensure that the results were representative. Details are given in Table 2. Six samples each were taken from sites with recuperative thickening, i.e., digester contents (digestate), recuperative thickener feed, recuperative thickened return, recuperative centrate, centrifuge dewatered cake, and centrifuge centrate in order to identify impact of both drum based thickening and centrifuge dewatering. All centrifuge dewatered and rotary drum thickened samples utilised cationic organic matrix polymer at industry standard dose rates (1e5 kg/Tonne DS). Three samples each were taken from sites without recuperative thickening, i.e., digester contents (digestate), centrifuge cake, and centrifuge centrate. All samples were transferred in cool boxes overnight, and held in 35 C incubators for 1e2 d prior to activity testing.
2.2.
Analysis
Standard methods were used for all basic chemical analysis (Franson et al., 2005). TS and VS were measured using crucible based drying (105 C) and volatilisation (550 C) method. COD was measured by dilution and analysis with a Merck 14543 spectroquant cell test. Organic acids were measured with a HP/Agilent GC-FID. Ammonia was measured using Ion Chromatography. Gas composition was measured using GCeTCD.
2.2.1.
Specific methanogenic activity
Specific methanogenic activity identifies the production of methane from acetate at above saturation kinetics and is analogous to biomass-specific maximum acetotrophic uptake rate (gCOD gVS1d1). Samples were degassed for 24e48 h in a 37 C incubator. 50 mL samples (undiluted for digestate or diluted for cake and thickened sludge) and 50 mL 2.0gCH3COOH (as CH3COONa) were placed in a 160 mL serum flask. Serum flasks were flushed using high purity nitrogen (99.9%) and immediately capped, and incubated at 37 C. All activity analyses were in triplicate. Methane was analysed periodically using gas displacement, and methane content analysed using Gas ChromatographyeThermal Conductivity Detection (GCeTCD). The system was a Perkin Elmer auto system GCeTCD with a 2.44 m stainless steel column packed with Haysep (80/100 mesh). The GC was fitted with a GC Plus
Data station, Model 1022 (Perkin Elmer, Waltham, MA, USA). High purity nitrogen (99.99%) was used as carrier gas at a flow rate of 24.3 mL min1 and a pressure of 220 kPa. The injection port, oven and detector were operated at 75 C, 40 C and 100 C, respectively. The GC was calibrated using external gas standards from British Oxygen Company (Sydney, Australia). Gas volume production and methane concentration were used to calculate methane produced. Triplicate analyses for each time were divided by inoculum biomass (estimated using VS), plotted against time, and the slope assessed using linear regression to provide the biomass specific methane production rate.
2.2.2.
DNA extraction and 16s pyrotag analysis
Digestate community only was analysed, on the basis that dewatering would only cause cell death or inactivation, not loss of DNA. Community genomic DNA from the sludge samples were extracted using FastDNA SPIN for Soil kit (MP Biomedicals, USA) and Fastprep beadbeating machine (Bio101, USA) according to manufacturer's protocol. The 3’ region of the 16S/18S rRNA gene was targeted using universal primers 926F (5’-AAACTYAAAKGAATTGACGG-3’) and 1392R (5’ACGGGCGGTGWGTRC-3’). Primer sequences were modified by the addition of Roche 454 adaptor 1 or 2 sequences and unique 5 bp barcodes at the 5’ end of the primer (sequences not shown but referenced as (Engelbrektson et al., 2010; Kunin et al., 2010)). DNA concentration and purity was then determined by gel electrophoresis on 1% agarose gel and spectrophotometrically using the NanoDrop ND-1000 (Thermo Fisher Scientific, USA). DNA was lyophilised using Savant SpeedVac Concentrator SVC100H (Thermo Fisher Scientific, USA) and submitted to the Australian Centre for Ecogenomics (ACE) for 16s rRNA gene pyrotag sequencing on the Genome Sequencer FLX Titanium platform (Roche, USA). Pyrotag sequences were processed using Pyrotagger (Kunin and Hugenholtz, 2010) and QIIME (Caporaso et al., 2010) with correction via ACACIA (Bragg et al., 2012).
2.3.
Statistical analysis and linear modelling
Error in replicates was based on a two-tailed t-test with n1 degrees of freedom (n ¼ number of replicates), and a/2 ¼ 0.025 (95% confidence interval). Specific methanogenic activity (SMA) was calculated based on the slope of cumulative methane produced per gVS inoculum, and error in SMA calculated from the standard error in slope (from regression)
w a t e r r e s e a r c h 8 2 ( 2 0 1 5 ) 7 8 e8 5
and a two-tailed t-test with n2 degrees of freedom (n ¼ number of points), and a/2 ¼ 0.025 (95% confidence interval). Error in decrease in relative activity was based on propagation principles given in Batstone (2013) for division of two variables. 95% confidence interval in linear model output was calculated based on the following formula: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 ðx xÞ2 þ E95;m Y ¼ ta=2;n2 s ^ n Sxx
(1)
Where ta/2,n2 is the critical two-tail t-test value for a/2 ¼ 0.025, s is the standard error in regression, n is the number of points in the regression, x is the mean of the regressor x, and Sxx is the variance of x. For microbial community analysis, OTU tables were normalised, and a square root transformation was applied to emphasise comparison of niche populations over dominants. A principal components analysis was then done on the square root, normalised OTU table using Matlab 2013b (princomp command), and results visualised using biplot. PC scores and PC variances were retrieved directly during use of the princomp command.
3.
Results
3.1.
Solids results
A summary of solids concentrations for the recuperative plants is given in Table 3 and non-recuperative plants in Table 4. This indicates that the rotary drum units used for recuperative thickening have a relatively low cake solids compared with final dewatering units as per their design. In all cases, centrate
Table 3 e Analysis of solids for recuperative plants. Site A
81
is very low in total solids concentration (indicating effective capture), and has a far higher fractional mineral solids, probably due to dissolved salts.
3.2.
Specific methanogenic activity (SMA)
SMA results returned highly linear sections with a short initial delay, as shown in Supplementary Info SII and could be estimated accurately for all samples with solids concentrations >1 g L1. The exception was reject (especially final centrifuges centrate), which generally could not be determined accurately, due to the low solids present in these streams (and hence have been removed from the analysis). Centrifuge based dewatering has a varied and strong impact on activity. Most plants are in the expected range of SMA of 0.2e0.4 gCOD gVS1d1 seen in other healthy digesters, but all display a substantial drop in SMA through centrifuge dewatering. The plant with highest sensitivity in SMA was Site A, followed by Site D. Impact varied from minor (site C), such that viability would be reasonably good, to major (site B), such that the residue would have minimal activity. Results from the recuperative thickening plants are shown in Fig. 2. This indicates that recuperative thickening (comparing feed and thickened activity) had no impact at Site A (i.e., no significant drop in activity), and a minor impact at site B (drop in activity of 20%). In all cases, activity of centrate was very difficult to measure due to the very low solids concentrations and hence has been removed.
3.3.
Microbial community
The microbial community in general was dominated by the acetoclastic methanogen Methanosaeta (average relative abundance 25%), with an OTU from uncultured candidate phyla WS6 being next most abundant (average abundance 8%). No OTUs affiliated with Methanosarcina were found with
Site B
TS (%) VS (%) VSfrac TS (%) VS (%) VSfrac Digester outlet Recup feed Recup return Recup centrate Final cake Final centrate
2.4 2.5 6.2 0.16 33.7 0.20
1.6 1.6 4.2 0.04 21.5 0.09
66% 64% 68% 28% 64% 43%
3.0 2.9 10.7 0.27 27.7 0.16
2.0 2.0 7.2 0.12 18.7 0.04
67% 68% 68% 45% 67% 25%
Concentration values % w/w.
Table 4 e Analysis of solids for non-recuperative plants. Digester outlet
Site A Site B Site C Site D Site E Site F Site G
Final cake
TS (%)
VS (%)
VSfrac
TS (%)
VS (%)
VSfrac
2.4 3.0 2.1 1.8 1.4 1.4 2.4
1.6 2.0 1.5 0.9 1.0 1.0 1.5
66% 67% 74% 54% 68% 66% 62%
33.7 27.7 17.8 28.8 19.3 25.8 26.3
21.5 18.7 13.6 16.7 13.8 17.8 16.5
64% 67% 76% 58% 71% 69% 63%
Concentration values % w/w.
Fig. 1 e Activity before (solid) and after (hatched) centrifuge based dewatering. Error bars indicate 95% confidence in average values from triplicate analyses.
82
w a t e r r e s e a r c h 8 2 ( 2 0 1 5 ) 7 8 e8 5
Fig. 2 e Activity before and after recuperative thickening including (for each site, left to right, digestate; final cake; final centrate; recup feed, recup return, and recup centrate). Blue bars indicate activity in feed to treatment, red bars indicate activity after treatment. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).
Fig. 3 e Principal Components analysis of the microbial population, indicating major vectors (blue) and samples (red) in the PCA space. Component 1 represents 36% of total variance, while Component 2 represents 25% of total variance. Sites A and B are with recuperative thickening. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article). all Methanosarcinales being Methanosaeta. PCA analysis of the microbial communities indicated that samples occupied one of three major spaces. Most of the standard digesters occupied the lower left quadrant, with a relatively balanced microbial
population (Fig. 3). PC1 described mainly shift from an acetoclastic community to hydrogen utilising methanogenic community (Methanobacterium and Methanomicrobiales). The two recuperative thickening plants occupied the right hand
w a t e r r e s e a r c h 8 2 ( 2 0 1 5 ) 7 8 e8 5
83
Fig. 4 e Impact of PC3 on apparent sensitivity. Correlation excludes outlier Site “C” which has a very low dewatering performance. Error bars indicate 95% confidence in calculated sensitivity.
hemisphere, with an increase in particularly hydrogen utilising methanogens, including Methanobacterium and Methanomicrobiales. Site E was in its own region, with a dominance of Methanosaeta (relative abundance 40%), and WS6, and a general decrease in diversity. Compared with nonrecuperative thickening plants, sites A and B demonstrate a strong difference (with a shift towards hydrogen utilising methanogens, p ¼ 0.0024 on PC1). As noted above, there is no significant difference in their specific methanogenic activity since sites A and B are in the median range in Fig. 1.
4.
Discussion
4.1.
Drivers of sensitivity to shear
Principal component 3 (PC3: 16% of variance), was an extremely strong predictor for the drop in methanogenic activity (p ¼ 0.0005 n ¼ 7). Analysing this in a scatter diagram (Fig. 4), there is a key outlier (Site C), and is an outlier in terms of having a very wet final cake compared to the others (18%). When this was excluded from the analysis, the significance increased (p ¼ 0.00001 n ¼ 6), and R2 increased to 0.997. The reason for the outlier is likely the low solids concentration in the centrifuge, with all other cake solids being >20%, except site E, which also had a low drop in activity. Analysing PC3 further, it is driven by shifts in population from one group to another with similar broad functionality (Fig. 5). For example, there is a shift from Methanosaeta and Methanobacterium in sensitive populations to Methanomicrobiales and Methanospirillum in less sensitive
populations. A number of organisms (such as Mollicutes) have a particularly low correlation to sensitivity. While sensitivity of bacterial populations have not been previously analysed, these results are supported when assessing membrane bioreactor systems subject to shear, with a shift from Methanobacteriales to Methanomicrobiales noted under shear (Padmasiri et al., 2007). Sensitivity of Methanosaeta to highshear conditions has also been widely noted (Kundu et al.,
Fig. 5 e PC3 score for specific OTUs.
84
w a t e r r e s e a r c h 8 2 ( 2 0 1 5 ) 7 8 e8 5
2014; Padmasiri et al., 2007), and we note that this normally is accompanied by a shift towards Methanosarcina. We hypothesise that (a) the shift is longer term, and more related to biofilm disruption than disruption at a cellular scale, and hence does not result in an immediate loss in activity, and (b) that the lack of Methanosarcina in general in these reactors means we are unable to test whether they are particularly sensitive. The relative importance of bacteria (e.g., Mollicutes), is also an interesting result, since even though methanogenesis is the key process, mediated by Archaea, it identifies flanking, correlated bacterial communities (e.g., Mollicutes), that may have a direct role enabling some capacity through acetate oxidation, or otherwise enabling shear resistant methanogenic archaea. Based on these results, we suggest that sensitivity is influenced by both solids induced shear as noted previously (Deveci, 2002; Kundu et al., 2014; Padmasiri et al., 2007) and as specifically identified by community (PC3), but we identify also that at very low solids (or shear), the link to microbial community is lost, and the material is insensitive simply due to the low applied shear (Site C).
4.2.
Microbial response to recuperative thickening
The microbial response to recuperative thickening was a long term response (all plants had operated with recuperative thickening for over 1 year), and was the strongest single driver of microbial community, with a shift towards hydrogen utilisers Methanobacteriales, Methanomicrobiales, and towards specific bacterial groups, particularly Clostridia from a more diversified bacterial community. Both of these were concentrated on key OTUs. Both shifts are also reasonable, since the shift following recuperative thickening is towards aerotolerant organisms within both domains. What was important was there was apparently no change in functionality as measured by specific methanogenic activity, or in sensitivity to shearing (as measured previously). Despite a deep literature search, no information could be found regarding broad microbial community shifts in anaerobic digesters in response to aeration or filter based thickening, and hence we lack perspective for the results found, which apparently indicate this is the single largest predictor of microbial community variation across the 7 sites studied.
4.3.
Significance of results
The results indicate that centrifuge, or other methods that expose the biomass to high levels of shear such as membrane based thickening (Ghyoot and Verstraete, 1997) may be unsuitable for recuperative thickening where recirculation rates are high. Particularly, methods such as lysis centrifuge, which focuses on increasing shear rates to increase degradability nyos et al., 1997) are particularly unsuitable for recu(Doha perative thickening (though they may be effective as pretreatment methods). Where centrifuge thickening method does cause deactivation, an increase in return rates will improve hydrolysis, and overall reactor performance (methane flow and VS destruction), but at the risk of methanogen deactivation. Above a critical return rate, excessive methanogen deactivation would result in acid failure. It is
possible to use high-shear methods for recuperative thickening, and piloting or simulations using (for example, the ADM1) could be used to determine a maximum acceptable recirculation rate given a per-pass deactivation. An initial model based analysis indicates that for primary sludge, a 20% deactivation is safe for recirculation rates <100% of external feed. Naturally, long term operation may further enable a shear resistant community as identified above. Given the apparent per-pass loss in methanogenic activity, the question remains as to whether there is justifiable benefit using a centrifuge for recuperative thickening over filter based methods (rotary drum or belt filter press). Centrifuges can achieve substantially higher solids compared to rotary drum thickeners, with a smaller footprint, at the expense of higher energy consumption (approx. 0.3 kWh kgDS1 for centrifuges vs 0.03 kWh kgDS1 for drum thickeners (WEF, 2010)). Taking practically a drum thickener operating at 8% vs a centrifuge at 12% (the limit at which sludge can still be effectively hydraulically pumped (Albertson et al., 1987)), three times the recirculation must be applied in a drum thickener vs a centrifuge to achieve the same performance (Supplementary information S1). Ultimately though, the limit at which recirculation rate can be applied is the operational limit of 5e6% indigester solids concentration, above which sludge rheology makes mixing very difficult due to high viscosity and increasingly non-Newtonian behaviour (Baudez et al., 2011; Tchobanoglous et al., 2003). This can be achieved at reasonable recirculation rates (~100% of feed), even with rotary drum thickeners. Overall, shear based thickening is less favourable for recuperative thickening given the marginal hydraulic gains, negative energy connotations, and decrease in activity. The final question is to whether centrifuge dewatered sludge can be used as inoculum to start up other digesters, which can offer substantial transport savings and logistical benefits over whole digestate. Given an average loss in activity of 54%, and only one dewatered sludge with unacceptably low activity (<0.03 gCOD gVS1d1), centrifuged anaerobic digestate generally can be used to inoculate new reactors.
5.
Conclusions
Dewatering through centrifuges caused a significant loss in specific methanogenic activity (averaging 54%), while rotary drum thickening caused a minor loss at one site (20%), and no loss at the other. While decreasing overall methanogenic capacity, this would not necessarily result in poorer performance where hydrolysis is rate-limiting (as hydrolysis may be improved), but it does impose a maximum return rate. Longterm recuperative thickening was the major driver for microbial population variance across the 7 sites, causing a shift from aceticlastic methanogens towards hydrogenotrophic methanogens. The major driver for sensitivity to shearing appeared to be microbial community (p ¼ 0.0005, or 0.00001 with outlier excluded). The one outlier to this relationship amongst the 7 sites examined had a high cake water content, possibly indicating it was exposed to a lower shear rate. Microbial community shifts in highly sensitive sites vs those less sensitive where mainly associated with a shift towards
w a t e r r e s e a r c h 8 2 ( 2 0 1 5 ) 7 8 e8 5
Methanomicrobiales and away from Methanobacteriales and Methanosaeta, but was also accompanied by a shift in bacterial flanking communities.
Acknowledgements Sydney Water funded this study and provided samples and operational data as well as technical review. Fiona May and the Australian Centre for Ecogenomics performed final amplification, tagging, and sequencing.
Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2015.04.005.
references
Albertson, O.E., Burris, B.E., Reed, S.C., Semon, J.A., Smith, J.E., Wallace, A.T., 1987. US EPA Design Manual: Dewatering Municipal Sludges. United States Environmental Protection Agency Office of Research and Development, Washington, DC. EPA/625/1e87/014. Batstone, D.J., 2013. Teaching uncertainty propagation as a core component in process engineering statistics. Educ. Chem. Eng. 8 (4), e132ee139. Baudez, J.C., Markis, F., Eshtiaghi, N., Slatter, P., 2011. The rheological behaviour of anaerobic digested sludge. Water Res. 45 (17), 5675e5680. Bragg, L., Stone, G., Imelfort, M., Hugenholtz, P., Tyson, G.W., 2012. Fast, accurate error-correction of amplicon pyrosequences using Acacia. Nat. Methods 9 (5), 425e426. Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K., Fierer, N., Pena, A.G., Goodrich, J.K., Gordon, J.I., Huttley, G.A., Kelley, S.T., Knights, D., Koenig, J.E., Ley, R.E., Lozupone, C.A., McDonald, D., Muegge, B.D., Pirrung, M., Reeder, J., Sevinsky, J.R., Turnbaugh, P.J., Walters, W.A., Widmann, J., Yatsunenko, T., Zaneveld, J., Knight, R., 2010. QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7 (5), 335e336. Chang, L.L., Raudenbush, D.L., Dentel, S.K., 2001. Aerobic and Anaerobic Biodegradability of a Flocculant Polymer, pp. 461e468. Conklin, A., Bucher, R., Stensel, H.D., Ferguson, J., 2007. Effects of oxygen exposure on anaerobic digester sludge. Water Environ. Res. 79 (4), 396e405.
85
Deveci, H., 2002. Effect of solids on viability of acidophilic bacteria. Miner. Eng. 15 (12), 1181e1189. nyos, M., Za branska , J., Jenı´cek, P., 1997. Enhancement of Doha sludge anaerobic digestion by using of a special thickening centrifuge. Water Sci. Technol. 36 (11), 145e153. Engelbrektson, A., Kunin, V., Wrighton, K.C., Zvenigorodsky, N., Chen, F., Ochman, H., Hugenholtz, P., 2010. Experimental factors affecting PCR-based estimates of microbial species richness and evenness. ISME J. 4 (5), 642e647. Franson, M.A.H., Eaton, A.D., Association, A.P.H., Association, A.W.W., Federation, W.E., 2005. Standard Methods for the Examination of Water & Wastewater. American Public Health Association, Washington, DC. Ghyoot, W.R., Verstraete, W.H., 1997. Coupling membrane filtration to anaerobic primary sludge digestion. Environ. Technol. 18 (6), 569e580. Kundu, K., Bergmann, I., Klocke, M., Sharma, S., Sreekrishnan, T.R., 2014. Influence of hydrodynamic shear on performance and microbial community structure of a hybrid anaerobic reactor. J. Chem. Technol. Biotechnol. 89 (3), 462e470. Kunin, V., Hugenholtz, P., 2010. PyroTagger: a fast, accurate pipeline for analysis of rRNA amplicon pyrosequence data. Open J. 1, 1e8. Kunin, V., Engelbrektson, A., Ochman, H., Hugenholtz, P., 2010. Wrinkles in the rare biosphere: pyrosequencing errors can lead to artificial inflation of diversity estimates. Environ. Microbiol. 12 (1), 118e123. Madigan, M.T., Madigan, M.T., Brock, T.D., 2009. Brock Biology of Microorganisms. Pearson Education, San Francisco, Calif. Muller, D.T., 2006. Shear Forces, Floc Structure and Their Impact on Anaerobic Digestion and Biosolids Stability. Virginia Polytech Institute, Blacksburg, Virginia. Padmasiri, S.I., Zhang, J., Fitch, M., Norddahl, B., Morgenroth, E., Raskin, L., 2007. Methanogenic population dynamics and performance of an anaerobic membrane bioreactor (AnMBR) treating swine manure under high shear conditions. Water Res. 41 (1), 134e144. Reynolds, D.T., Cannon, M., Pelton, T., 2001. Preliminary investigation of recuperative thickening for anaerobic digestion. Proc. Water Environ. Fed. 2001 (14), 389e410. Tchobanoglous, G., Burton, F., Stensel, H., 2003. Metcalf and Eddy Inc. Wastewater Engineering, Treatment and Reuse. McGrawHill., New York, NY (US). Vanyushina, A.Y., Agarev, A.M., Moyzhes, S.I., Nikolaev, Y.A., Kevbrina, M.V., Kozlov, M.N., 2012. Comparison of different thickening methods for active biomass recycle for anaerobic digestion of wastewater sludge. Water Sci. Technol. 66 (8), 1787e1793. WEF, 2010. Energy Conservation in Water and Wastewater Treatment Facilities/Prepared by the Energy Conservation in Water and Wastewater Treatment Facilities Task Force of the Water Environment Federation, WEF Press, Water Environment Federation; New York. McGraw Hill, Alexandria, VA.
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/watres
Impact of dewatering technologies on specific methanogenic activity Damien J. Batstone*, Yang Lu, Paul D. Jensen Advanced Water Management Centre (AWMC), The University of Queensland, St Lucia, 4072, Brisbane, Queensland, Australia
article info
abstract
Article history:
Dewatering methods for recuperative thickening and final dewatering can potentially
Received 30 November 2014
impact methanogenic activity and microbial community. This influences both the feasi-
Received in revised form
bility of recuperative thickening to increase solids residence time within a digester, and the
20 March 2015
utilisation of dewatered digestate as inoculum for new digesters. Thickening technology
Accepted 2 April 2015
can reduce methanogenic activity through either air contact (rotary drum, DAF, or belt filter press), or by lysing cells through shear (centrifuge). To assess this, two plants with recuperative thickening (rotary drum) in their anaerobic digester, and five without recu-
Keywords:
perative thickening, had specific methanogenic activity tested in all related streams,
Recuperative
including dewatering feed, thickened return, final cake, and centrate. All plants had high
Thickening
speed centrifuges for final dewatering. The digester microbial community was also
Shear
assessed through 16s pyrotag sequencing and subsequent principal component analysis
Specific methanogenic activity
(PCA). The specific methanogenic activity of all samples was in the expected range of 0.2
Pyrotag sequencing
e0.4 gCOD gVS1d1. Plants with recuperative thickening did not have lower digester activity. Centrifuge based dewatering had a significant and variable impact on methanogenic activity in all samples, ranging between 20% and 90% decrease but averaging 54%. Rotary drum based recuperative thickening had a far smaller impact on activity, with a 0% perpass drop in activity in one plant, and a 20% drop in another. However, the presence of recuperative thickening was a major predictor of overall microbial community (PC1, p ¼ 0.0024). Microbial community PC3 (mainly driven by a shift in methanogens) was a strong predictor for sensitivity in activity to shear (p ¼ 0.0005, p ¼ 0.00001 without outlier). The one outlier was related to a plant producing the wettest cake (17% solids). This indicates that high solids is a potential driver of sensitivity to shear, but that a resilient microbial community can also bestow resilience. Sensitivity of methanogens to centrifuging does not rule out centrifuges for recuperative thickening (particularly where hydrolysis is rate-limiting), but may impose a maximum return rate to avoid digester failure. © 2015 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: þ61 3346 9051; fax: þ61 7 3365 4726. E-mail address: [email protected] (D.J. Batstone). http://dx.doi.org/10.1016/j.watres.2015.04.005 0043-1354/© 2015 Elsevier Ltd. All rights reserved.
w a t e r r e s e a r c h 8 2 ( 2 0 1 5 ) 7 8 e8 5
1.
Introduction
Recuperative thickening is a process for intensifying anaerobic digesters, by removing water from digestate, and returning the high-solids concentrate back to the digester. It thereby increases effective solids retention time over hydraulic retention time (Reynolds et al., 2001; Vanyushina et al., 2012). It is particularly suited to primary sludge, which has a relatively high degradability extent, but which is also relatively slow to degrade, and generally poorly suited material with a limited degradability extent, since this will result in the recycling of large amounts of inert material (see model provided supplementary information SI). Recuperative thickening is a relatively low cost technique for intensification, with well understood fundamentals, but there is very references found in the formal literature. A key practical concern is the possibility of damaging anaerobic microbes, particularly methanogenic archaea during the thickening stage. The impact of recuperative thickening on methanogenic activity is of critical importance, since the digester contents may be passed through the thickening equipment multiple times before it leaves the digester for final dewatering. The other major process of water removal from sludge is final dewatering, and indeed, centrifuges can be used for both recuperative thickening and final dewatering. The impact of final dewatering on activity is also important since firstly, dewatered sludge may be used as inoculum in other digesters, and secondly, the activity of dewatered material potentially determines methane production during storage. Both recuperative thickening and final dewatering can be regarded as comparable processes. In the case of recuperative thickening, digester sludge is thickened to 6%e12% through a range of techniques, including centrifuges and belt filter presses (Tchobanoglous et al., 2003), dissolved air flotation (Vanyushina et al., 2012), gravity belt thickeners (Reynolds et al., 2001), and rotary drum thickeners (Tchobanoglous et al., 2003). These can be generally classed by the DS achieved and the potential impact on the biology as shown in Table 1. Final dewatering depends on high-g and standard centrifuges, or belt filter presses to produce a final cake at >12% DS (Albertson et al., 1987) whereas recuperative thickening technologies target lower DS. The major factors that may influence activity are liquid shear and oxygen exposure (Table 1). Flocculant is not inhibitory at normal doses, and indeed, may partially degrade
Table 1 e Dewatering and thickening equipment summary. Equipment High-g centrifuge Centrifuge Belt filter press Rotary drum Baleen filter Dissolved air flotation Gravity belt thickening
Performance (% DS)
Biochemical impact
18%e30% 15%e25% 12%e25% 8%e15% 5%e8% 5%e8% 4%e6%
High shear Shear Oxygen, mild shear Oxygen Oxygen, mild shear High oxygen Oxygen
From Albertson et al. (1987), Tchobanoglous et al. (2003).
79
(Chang et al., 2001). Liquid shear occurring at impeller tip velocities >2 m s1 (shear rate e G~2000 s1 in a 1 mm tip field) are of concern since this lyse susceptible organisms, and may disrupt microbial syntrophic associations (Deveci, 2002; Ghyoot and Verstraete, 1997), while oxygen is toxic to methanogenic archaea in particular (Madigan et al., 2009). High-g equipment such as centrifuges (G 5000 s1 (Muller, 2006))are more likely to expose organisms to shear, while dissolved air filtration, and rotary drum thickeners in particular are more likely to expose organisms to oxygen (Table 1). The issue of oxygen exposure during thickening for recuperative purposes has been previously considered. Conklin et al. (2007) (Conklin et al., 2007) assessed the impact of oxygen exposure (via sparging) as a proxy for dissolved air flotation and actual samples from a gravity belt thickener (GBT) to assess the potential use of either for recuperative thickening. They found (via activity) that mild exposure (i.e., GBT) had no impact on activity while high exposure (i.e., sparging) only had an impact, but would only be of effect at high relative recirculation rates, or where the reactor was overloaded. The impact of centrifuge dewatering on activity has not been previously considered, either for recuperative thickening, or for final dewatering, and particularly, the impact of shear on methanogens has not been investigated. However, Deveci (2002) found that shearing had a strong impact on viability of aerobic acidophilic organisms particularly at solids concentrations above 10%. The finding was that shear alone was less likely to result in loss of activity, but that at high solids, poor momentum dissipation, and hence a high velocity gradient (i.e., shear) would result in loss of activity. Application of membrane based methods to selectively remove permeate have consistently found reduced activities at higher recirculation rates (Ghyoot and Verstraete, 1997), and high shear in a membrane configuration is correlated with a loss of acetoclastic methanogens (Padmasiri et al., 2007). As seen in Table 1, particularly mechanical dewatering equipment operates in an environment of high-shear and high solids, and the impact of dewatering equipment on anaerobes and particularly methanogens is unknown. While there is relevant incidental literature, there is very little available on the impact of dewatering on anaerobic systems, and particularly on activity, more so for the purpose of recuperative thickening. This is surprising given the relatively feasibility and simple application of recuperative thickening to intensify anaerobic processes. This paper addresses the key gap of impact of shear (centrifuge) based dewatering and air based (rotary drum) thickening on methanogenic activity, and also assesses the impact of long-term operation using rotary drum thickening on microbial community.
2.
Methods
2.1.
Digesters and sampling
27 samples were taken from 7 sites in the Sydney greater metropolitan area (plant names given in Table 2). These included two sites with recuperative thickening (A, B e six samples each), and five sites without recuperative thickening
80
w a t e r r e s e a r c h 8 2 ( 2 0 1 5 ) 7 8 e8 5
Table 2 e Treatment processes in the various plants studied. Ident A B C D E F G a b
Site North head Bondi Cronulla Malabar Glenfield Liverpool Wollongong
Feed to digester Primary Primary Blended Primary Blended Blended Blended
Loading kgVS/m3 d1
Digester HRT (d)
Recup
3.0 2.3 3.4 2.6 1.8 1.8 2.7
13 10 10 12 17 19 13.5
RDTb RDT N/A N/A N/A N/A N/A
(70% primary)a (70% primary) (70% primary) (60% primary)
Indicates load by mass dry solids. All plants use high-speed centrifuges for final dewatering; RDT ¼ rotary drum thickener.
(C, D, E, F, G e 3 samples each). Sites A and B used rotary drum recuperative thickening, and high-g centrifuges for final dewatering, while sites C-G used high-g centrifuges for final dewatering. Solids analysis results were compared with historical data to ensure that the results were representative. Details are given in Table 2. Six samples each were taken from sites with recuperative thickening, i.e., digester contents (digestate), recuperative thickener feed, recuperative thickened return, recuperative centrate, centrifuge dewatered cake, and centrifuge centrate in order to identify impact of both drum based thickening and centrifuge dewatering. All centrifuge dewatered and rotary drum thickened samples utilised cationic organic matrix polymer at industry standard dose rates (1e5 kg/Tonne DS). Three samples each were taken from sites without recuperative thickening, i.e., digester contents (digestate), centrifuge cake, and centrifuge centrate. All samples were transferred in cool boxes overnight, and held in 35 C incubators for 1e2 d prior to activity testing.
2.2.
Analysis
Standard methods were used for all basic chemical analysis (Franson et al., 2005). TS and VS were measured using crucible based drying (105 C) and volatilisation (550 C) method. COD was measured by dilution and analysis with a Merck 14543 spectroquant cell test. Organic acids were measured with a HP/Agilent GC-FID. Ammonia was measured using Ion Chromatography. Gas composition was measured using GCeTCD.
2.2.1.
Specific methanogenic activity
Specific methanogenic activity identifies the production of methane from acetate at above saturation kinetics and is analogous to biomass-specific maximum acetotrophic uptake rate (gCOD gVS1d1). Samples were degassed for 24e48 h in a 37 C incubator. 50 mL samples (undiluted for digestate or diluted for cake and thickened sludge) and 50 mL 2.0gCH3COOH (as CH3COONa) were placed in a 160 mL serum flask. Serum flasks were flushed using high purity nitrogen (99.9%) and immediately capped, and incubated at 37 C. All activity analyses were in triplicate. Methane was analysed periodically using gas displacement, and methane content analysed using Gas ChromatographyeThermal Conductivity Detection (GCeTCD). The system was a Perkin Elmer auto system GCeTCD with a 2.44 m stainless steel column packed with Haysep (80/100 mesh). The GC was fitted with a GC Plus
Data station, Model 1022 (Perkin Elmer, Waltham, MA, USA). High purity nitrogen (99.99%) was used as carrier gas at a flow rate of 24.3 mL min1 and a pressure of 220 kPa. The injection port, oven and detector were operated at 75 C, 40 C and 100 C, respectively. The GC was calibrated using external gas standards from British Oxygen Company (Sydney, Australia). Gas volume production and methane concentration were used to calculate methane produced. Triplicate analyses for each time were divided by inoculum biomass (estimated using VS), plotted against time, and the slope assessed using linear regression to provide the biomass specific methane production rate.
2.2.2.
DNA extraction and 16s pyrotag analysis
Digestate community only was analysed, on the basis that dewatering would only cause cell death or inactivation, not loss of DNA. Community genomic DNA from the sludge samples were extracted using FastDNA SPIN for Soil kit (MP Biomedicals, USA) and Fastprep beadbeating machine (Bio101, USA) according to manufacturer's protocol. The 3’ region of the 16S/18S rRNA gene was targeted using universal primers 926F (5’-AAACTYAAAKGAATTGACGG-3’) and 1392R (5’ACGGGCGGTGWGTRC-3’). Primer sequences were modified by the addition of Roche 454 adaptor 1 or 2 sequences and unique 5 bp barcodes at the 5’ end of the primer (sequences not shown but referenced as (Engelbrektson et al., 2010; Kunin et al., 2010)). DNA concentration and purity was then determined by gel electrophoresis on 1% agarose gel and spectrophotometrically using the NanoDrop ND-1000 (Thermo Fisher Scientific, USA). DNA was lyophilised using Savant SpeedVac Concentrator SVC100H (Thermo Fisher Scientific, USA) and submitted to the Australian Centre for Ecogenomics (ACE) for 16s rRNA gene pyrotag sequencing on the Genome Sequencer FLX Titanium platform (Roche, USA). Pyrotag sequences were processed using Pyrotagger (Kunin and Hugenholtz, 2010) and QIIME (Caporaso et al., 2010) with correction via ACACIA (Bragg et al., 2012).
2.3.
Statistical analysis and linear modelling
Error in replicates was based on a two-tailed t-test with n1 degrees of freedom (n ¼ number of replicates), and a/2 ¼ 0.025 (95% confidence interval). Specific methanogenic activity (SMA) was calculated based on the slope of cumulative methane produced per gVS inoculum, and error in SMA calculated from the standard error in slope (from regression)
w a t e r r e s e a r c h 8 2 ( 2 0 1 5 ) 7 8 e8 5
and a two-tailed t-test with n2 degrees of freedom (n ¼ number of points), and a/2 ¼ 0.025 (95% confidence interval). Error in decrease in relative activity was based on propagation principles given in Batstone (2013) for division of two variables. 95% confidence interval in linear model output was calculated based on the following formula: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 ðx xÞ2 þ E95;m Y ¼ ta=2;n2 s ^ n Sxx
(1)
Where ta/2,n2 is the critical two-tail t-test value for a/2 ¼ 0.025, s is the standard error in regression, n is the number of points in the regression, x is the mean of the regressor x, and Sxx is the variance of x. For microbial community analysis, OTU tables were normalised, and a square root transformation was applied to emphasise comparison of niche populations over dominants. A principal components analysis was then done on the square root, normalised OTU table using Matlab 2013b (princomp command), and results visualised using biplot. PC scores and PC variances were retrieved directly during use of the princomp command.
3.
Results
3.1.
Solids results
A summary of solids concentrations for the recuperative plants is given in Table 3 and non-recuperative plants in Table 4. This indicates that the rotary drum units used for recuperative thickening have a relatively low cake solids compared with final dewatering units as per their design. In all cases, centrate
Table 3 e Analysis of solids for recuperative plants. Site A
81
is very low in total solids concentration (indicating effective capture), and has a far higher fractional mineral solids, probably due to dissolved salts.
3.2.
Specific methanogenic activity (SMA)
SMA results returned highly linear sections with a short initial delay, as shown in Supplementary Info SII and could be estimated accurately for all samples with solids concentrations >1 g L1. The exception was reject (especially final centrifuges centrate), which generally could not be determined accurately, due to the low solids present in these streams (and hence have been removed from the analysis). Centrifuge based dewatering has a varied and strong impact on activity. Most plants are in the expected range of SMA of 0.2e0.4 gCOD gVS1d1 seen in other healthy digesters, but all display a substantial drop in SMA through centrifuge dewatering. The plant with highest sensitivity in SMA was Site A, followed by Site D. Impact varied from minor (site C), such that viability would be reasonably good, to major (site B), such that the residue would have minimal activity. Results from the recuperative thickening plants are shown in Fig. 2. This indicates that recuperative thickening (comparing feed and thickened activity) had no impact at Site A (i.e., no significant drop in activity), and a minor impact at site B (drop in activity of 20%). In all cases, activity of centrate was very difficult to measure due to the very low solids concentrations and hence has been removed.
3.3.
Microbial community
The microbial community in general was dominated by the acetoclastic methanogen Methanosaeta (average relative abundance 25%), with an OTU from uncultured candidate phyla WS6 being next most abundant (average abundance 8%). No OTUs affiliated with Methanosarcina were found with
Site B
TS (%) VS (%) VSfrac TS (%) VS (%) VSfrac Digester outlet Recup feed Recup return Recup centrate Final cake Final centrate
2.4 2.5 6.2 0.16 33.7 0.20
1.6 1.6 4.2 0.04 21.5 0.09
66% 64% 68% 28% 64% 43%
3.0 2.9 10.7 0.27 27.7 0.16
2.0 2.0 7.2 0.12 18.7 0.04
67% 68% 68% 45% 67% 25%
Concentration values % w/w.
Table 4 e Analysis of solids for non-recuperative plants. Digester outlet
Site A Site B Site C Site D Site E Site F Site G
Final cake
TS (%)
VS (%)
VSfrac
TS (%)
VS (%)
VSfrac
2.4 3.0 2.1 1.8 1.4 1.4 2.4
1.6 2.0 1.5 0.9 1.0 1.0 1.5
66% 67% 74% 54% 68% 66% 62%
33.7 27.7 17.8 28.8 19.3 25.8 26.3
21.5 18.7 13.6 16.7 13.8 17.8 16.5
64% 67% 76% 58% 71% 69% 63%
Concentration values % w/w.
Fig. 1 e Activity before (solid) and after (hatched) centrifuge based dewatering. Error bars indicate 95% confidence in average values from triplicate analyses.
82
w a t e r r e s e a r c h 8 2 ( 2 0 1 5 ) 7 8 e8 5
Fig. 2 e Activity before and after recuperative thickening including (for each site, left to right, digestate; final cake; final centrate; recup feed, recup return, and recup centrate). Blue bars indicate activity in feed to treatment, red bars indicate activity after treatment. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).
Fig. 3 e Principal Components analysis of the microbial population, indicating major vectors (blue) and samples (red) in the PCA space. Component 1 represents 36% of total variance, while Component 2 represents 25% of total variance. Sites A and B are with recuperative thickening. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article). all Methanosarcinales being Methanosaeta. PCA analysis of the microbial communities indicated that samples occupied one of three major spaces. Most of the standard digesters occupied the lower left quadrant, with a relatively balanced microbial
population (Fig. 3). PC1 described mainly shift from an acetoclastic community to hydrogen utilising methanogenic community (Methanobacterium and Methanomicrobiales). The two recuperative thickening plants occupied the right hand
w a t e r r e s e a r c h 8 2 ( 2 0 1 5 ) 7 8 e8 5
83
Fig. 4 e Impact of PC3 on apparent sensitivity. Correlation excludes outlier Site “C” which has a very low dewatering performance. Error bars indicate 95% confidence in calculated sensitivity.
hemisphere, with an increase in particularly hydrogen utilising methanogens, including Methanobacterium and Methanomicrobiales. Site E was in its own region, with a dominance of Methanosaeta (relative abundance 40%), and WS6, and a general decrease in diversity. Compared with nonrecuperative thickening plants, sites A and B demonstrate a strong difference (with a shift towards hydrogen utilising methanogens, p ¼ 0.0024 on PC1). As noted above, there is no significant difference in their specific methanogenic activity since sites A and B are in the median range in Fig. 1.
4.
Discussion
4.1.
Drivers of sensitivity to shear
Principal component 3 (PC3: 16% of variance), was an extremely strong predictor for the drop in methanogenic activity (p ¼ 0.0005 n ¼ 7). Analysing this in a scatter diagram (Fig. 4), there is a key outlier (Site C), and is an outlier in terms of having a very wet final cake compared to the others (18%). When this was excluded from the analysis, the significance increased (p ¼ 0.00001 n ¼ 6), and R2 increased to 0.997. The reason for the outlier is likely the low solids concentration in the centrifuge, with all other cake solids being >20%, except site E, which also had a low drop in activity. Analysing PC3 further, it is driven by shifts in population from one group to another with similar broad functionality (Fig. 5). For example, there is a shift from Methanosaeta and Methanobacterium in sensitive populations to Methanomicrobiales and Methanospirillum in less sensitive
populations. A number of organisms (such as Mollicutes) have a particularly low correlation to sensitivity. While sensitivity of bacterial populations have not been previously analysed, these results are supported when assessing membrane bioreactor systems subject to shear, with a shift from Methanobacteriales to Methanomicrobiales noted under shear (Padmasiri et al., 2007). Sensitivity of Methanosaeta to highshear conditions has also been widely noted (Kundu et al.,
Fig. 5 e PC3 score for specific OTUs.
84
w a t e r r e s e a r c h 8 2 ( 2 0 1 5 ) 7 8 e8 5
2014; Padmasiri et al., 2007), and we note that this normally is accompanied by a shift towards Methanosarcina. We hypothesise that (a) the shift is longer term, and more related to biofilm disruption than disruption at a cellular scale, and hence does not result in an immediate loss in activity, and (b) that the lack of Methanosarcina in general in these reactors means we are unable to test whether they are particularly sensitive. The relative importance of bacteria (e.g., Mollicutes), is also an interesting result, since even though methanogenesis is the key process, mediated by Archaea, it identifies flanking, correlated bacterial communities (e.g., Mollicutes), that may have a direct role enabling some capacity through acetate oxidation, or otherwise enabling shear resistant methanogenic archaea. Based on these results, we suggest that sensitivity is influenced by both solids induced shear as noted previously (Deveci, 2002; Kundu et al., 2014; Padmasiri et al., 2007) and as specifically identified by community (PC3), but we identify also that at very low solids (or shear), the link to microbial community is lost, and the material is insensitive simply due to the low applied shear (Site C).
4.2.
Microbial response to recuperative thickening
The microbial response to recuperative thickening was a long term response (all plants had operated with recuperative thickening for over 1 year), and was the strongest single driver of microbial community, with a shift towards hydrogen utilisers Methanobacteriales, Methanomicrobiales, and towards specific bacterial groups, particularly Clostridia from a more diversified bacterial community. Both of these were concentrated on key OTUs. Both shifts are also reasonable, since the shift following recuperative thickening is towards aerotolerant organisms within both domains. What was important was there was apparently no change in functionality as measured by specific methanogenic activity, or in sensitivity to shearing (as measured previously). Despite a deep literature search, no information could be found regarding broad microbial community shifts in anaerobic digesters in response to aeration or filter based thickening, and hence we lack perspective for the results found, which apparently indicate this is the single largest predictor of microbial community variation across the 7 sites studied.
4.3.
Significance of results
The results indicate that centrifuge, or other methods that expose the biomass to high levels of shear such as membrane based thickening (Ghyoot and Verstraete, 1997) may be unsuitable for recuperative thickening where recirculation rates are high. Particularly, methods such as lysis centrifuge, which focuses on increasing shear rates to increase degradability nyos et al., 1997) are particularly unsuitable for recu(Doha perative thickening (though they may be effective as pretreatment methods). Where centrifuge thickening method does cause deactivation, an increase in return rates will improve hydrolysis, and overall reactor performance (methane flow and VS destruction), but at the risk of methanogen deactivation. Above a critical return rate, excessive methanogen deactivation would result in acid failure. It is
possible to use high-shear methods for recuperative thickening, and piloting or simulations using (for example, the ADM1) could be used to determine a maximum acceptable recirculation rate given a per-pass deactivation. An initial model based analysis indicates that for primary sludge, a 20% deactivation is safe for recirculation rates <100% of external feed. Naturally, long term operation may further enable a shear resistant community as identified above. Given the apparent per-pass loss in methanogenic activity, the question remains as to whether there is justifiable benefit using a centrifuge for recuperative thickening over filter based methods (rotary drum or belt filter press). Centrifuges can achieve substantially higher solids compared to rotary drum thickeners, with a smaller footprint, at the expense of higher energy consumption (approx. 0.3 kWh kgDS1 for centrifuges vs 0.03 kWh kgDS1 for drum thickeners (WEF, 2010)). Taking practically a drum thickener operating at 8% vs a centrifuge at 12% (the limit at which sludge can still be effectively hydraulically pumped (Albertson et al., 1987)), three times the recirculation must be applied in a drum thickener vs a centrifuge to achieve the same performance (Supplementary information S1). Ultimately though, the limit at which recirculation rate can be applied is the operational limit of 5e6% indigester solids concentration, above which sludge rheology makes mixing very difficult due to high viscosity and increasingly non-Newtonian behaviour (Baudez et al., 2011; Tchobanoglous et al., 2003). This can be achieved at reasonable recirculation rates (~100% of feed), even with rotary drum thickeners. Overall, shear based thickening is less favourable for recuperative thickening given the marginal hydraulic gains, negative energy connotations, and decrease in activity. The final question is to whether centrifuge dewatered sludge can be used as inoculum to start up other digesters, which can offer substantial transport savings and logistical benefits over whole digestate. Given an average loss in activity of 54%, and only one dewatered sludge with unacceptably low activity (<0.03 gCOD gVS1d1), centrifuged anaerobic digestate generally can be used to inoculate new reactors.
5.
Conclusions
Dewatering through centrifuges caused a significant loss in specific methanogenic activity (averaging 54%), while rotary drum thickening caused a minor loss at one site (20%), and no loss at the other. While decreasing overall methanogenic capacity, this would not necessarily result in poorer performance where hydrolysis is rate-limiting (as hydrolysis may be improved), but it does impose a maximum return rate. Longterm recuperative thickening was the major driver for microbial population variance across the 7 sites, causing a shift from aceticlastic methanogens towards hydrogenotrophic methanogens. The major driver for sensitivity to shearing appeared to be microbial community (p ¼ 0.0005, or 0.00001 with outlier excluded). The one outlier to this relationship amongst the 7 sites examined had a high cake water content, possibly indicating it was exposed to a lower shear rate. Microbial community shifts in highly sensitive sites vs those less sensitive where mainly associated with a shift towards
w a t e r r e s e a r c h 8 2 ( 2 0 1 5 ) 7 8 e8 5
Methanomicrobiales and away from Methanobacteriales and Methanosaeta, but was also accompanied by a shift in bacterial flanking communities.
Acknowledgements Sydney Water funded this study and provided samples and operational data as well as technical review. Fiona May and the Australian Centre for Ecogenomics performed final amplification, tagging, and sequencing.
Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2015.04.005.
references
Albertson, O.E., Burris, B.E., Reed, S.C., Semon, J.A., Smith, J.E., Wallace, A.T., 1987. US EPA Design Manual: Dewatering Municipal Sludges. United States Environmental Protection Agency Office of Research and Development, Washington, DC. EPA/625/1e87/014. Batstone, D.J., 2013. Teaching uncertainty propagation as a core component in process engineering statistics. Educ. Chem. Eng. 8 (4), e132ee139. Baudez, J.C., Markis, F., Eshtiaghi, N., Slatter, P., 2011. The rheological behaviour of anaerobic digested sludge. Water Res. 45 (17), 5675e5680. Bragg, L., Stone, G., Imelfort, M., Hugenholtz, P., Tyson, G.W., 2012. Fast, accurate error-correction of amplicon pyrosequences using Acacia. Nat. Methods 9 (5), 425e426. Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K., Fierer, N., Pena, A.G., Goodrich, J.K., Gordon, J.I., Huttley, G.A., Kelley, S.T., Knights, D., Koenig, J.E., Ley, R.E., Lozupone, C.A., McDonald, D., Muegge, B.D., Pirrung, M., Reeder, J., Sevinsky, J.R., Turnbaugh, P.J., Walters, W.A., Widmann, J., Yatsunenko, T., Zaneveld, J., Knight, R., 2010. QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7 (5), 335e336. Chang, L.L., Raudenbush, D.L., Dentel, S.K., 2001. Aerobic and Anaerobic Biodegradability of a Flocculant Polymer, pp. 461e468. Conklin, A., Bucher, R., Stensel, H.D., Ferguson, J., 2007. Effects of oxygen exposure on anaerobic digester sludge. Water Environ. Res. 79 (4), 396e405.
85
Deveci, H., 2002. Effect of solids on viability of acidophilic bacteria. Miner. Eng. 15 (12), 1181e1189. nyos, M., Za branska , J., Jenı´cek, P., 1997. Enhancement of Doha sludge anaerobic digestion by using of a special thickening centrifuge. Water Sci. Technol. 36 (11), 145e153. Engelbrektson, A., Kunin, V., Wrighton, K.C., Zvenigorodsky, N., Chen, F., Ochman, H., Hugenholtz, P., 2010. Experimental factors affecting PCR-based estimates of microbial species richness and evenness. ISME J. 4 (5), 642e647. Franson, M.A.H., Eaton, A.D., Association, A.P.H., Association, A.W.W., Federation, W.E., 2005. Standard Methods for the Examination of Water & Wastewater. American Public Health Association, Washington, DC. Ghyoot, W.R., Verstraete, W.H., 1997. Coupling membrane filtration to anaerobic primary sludge digestion. Environ. Technol. 18 (6), 569e580. Kundu, K., Bergmann, I., Klocke, M., Sharma, S., Sreekrishnan, T.R., 2014. Influence of hydrodynamic shear on performance and microbial community structure of a hybrid anaerobic reactor. J. Chem. Technol. Biotechnol. 89 (3), 462e470. Kunin, V., Hugenholtz, P., 2010. PyroTagger: a fast, accurate pipeline for analysis of rRNA amplicon pyrosequence data. Open J. 1, 1e8. Kunin, V., Engelbrektson, A., Ochman, H., Hugenholtz, P., 2010. Wrinkles in the rare biosphere: pyrosequencing errors can lead to artificial inflation of diversity estimates. Environ. Microbiol. 12 (1), 118e123. Madigan, M.T., Madigan, M.T., Brock, T.D., 2009. Brock Biology of Microorganisms. Pearson Education, San Francisco, Calif. Muller, D.T., 2006. Shear Forces, Floc Structure and Their Impact on Anaerobic Digestion and Biosolids Stability. Virginia Polytech Institute, Blacksburg, Virginia. Padmasiri, S.I., Zhang, J., Fitch, M., Norddahl, B., Morgenroth, E., Raskin, L., 2007. Methanogenic population dynamics and performance of an anaerobic membrane bioreactor (AnMBR) treating swine manure under high shear conditions. Water Res. 41 (1), 134e144. Reynolds, D.T., Cannon, M., Pelton, T., 2001. Preliminary investigation of recuperative thickening for anaerobic digestion. Proc. Water Environ. Fed. 2001 (14), 389e410. Tchobanoglous, G., Burton, F., Stensel, H., 2003. Metcalf and Eddy Inc. Wastewater Engineering, Treatment and Reuse. McGrawHill., New York, NY (US). Vanyushina, A.Y., Agarev, A.M., Moyzhes, S.I., Nikolaev, Y.A., Kevbrina, M.V., Kozlov, M.N., 2012. Comparison of different thickening methods for active biomass recycle for anaerobic digestion of wastewater sludge. Water Sci. Technol. 66 (8), 1787e1793. WEF, 2010. Energy Conservation in Water and Wastewater Treatment Facilities/Prepared by the Energy Conservation in Water and Wastewater Treatment Facilities Task Force of the Water Environment Federation, WEF Press, Water Environment Federation; New York. McGraw Hill, Alexandria, VA.