Effect of SBR feeding strategy and feed composition on the stability of aerobic granular sludge in the treatment of a simulated textile wastewater

High organic loads, recalcitrant dyes and potentially toxic azo dye breakdown products have long been considered the top environmental issues associated with textile industry wastewater (O'Neill et al. 1999). The need for efficient and environmentally friendly dye removal processes led to the study of biological wastewater treatment, as an alternative to physical and/or chemical processes, which are expensive and create secondary pollution problems (dos Santos et al. 2007). The ability of bacteria to decolorize azo dyes (the dominant textile dye type) via anaerobic reduction, and to aerobically mineralize some of the resulting aromatic amines, motivated the application of anaerobic–aerobic sequencing batch reactors (SBRs) to treat textile wastewater (Sarayu & Sandhya 2012). Nevertheless, despite the success reported for the anaerobic decolorization stage (Lourenço et al. 2000), data on the fate of the aromatic amines during the aerobic stage, when available, revealed that most were not degraded (Van der Zee & Villaverde 2005). In this context, the novel aerobic granular sludge (AGS) (Beun et al. 1999; Pronk et al. 2015b), known for its high settleability, its resistance to toxic compounds and its potential for the co-existence of aerobic/anoxic microenvironments within granules, has been recently applied in studies with simulated textile wastewater (Muda et al. 2010; Franca et al. 2015).

Despite the advances in AGS technology, long-term stability of aerobic granules (AG), a key factor for stable operation of AGS SBRs, is still a challenge (Lee et al. 2010; Zhang et al. 2016). A low microbial growth rate inside AG has been described as an important factor to keep AG stability (de Kreuk & van Loosdrecht 2004). Regarding the SBR hydrodynamic and aeration regimens, anaerobic plug-flow feeding followed by aerobic reaction has been associated with improved structural stability of AGS in tubular SBRs being attributed to the selection of heterotrophic microorganisms able to anaerobically convert easily biodegradable substrates to slowly biodegradable storage polymers, thus exhibiting lower growth rates in the aerobic phase (de Kreuk & van Loosdrecht 2004; Pronk et al. 2015a). On the other hand, when comparing with the use of an anaerobic plug-flow feeding, a stirred anaerobic phase following a fast feeding period has been suggested to improve anaerobic substrate uptake prior to the aerobic stage in small height-to-diameter ratio (H/D) reactors (non-tubular), and consequently enhance AGS stability (Rocktäschel et al. 2013).

The reaction stage of both SBRs used in the present work (small H/D) included an anaerobic stirred phase followed by an aerobic stage, the difference between the reactors lying in the feeding strategy. Specifically, this study assessed the effect of two feeding regimens – an anaerobic plug-flow feeding through the sludge bed and a static anaerobic feeding – on the stability of AGS in non-tubular SBRs treating a synthetic textile wastewater, namely when subjected to changes in feed composition, with a focus on azo dye biodegradation and chemical oxygen demand (COD) removal.

Sodium acetate (analytical grade, AppliChem GmbH, Germany) and/or a starch-based sizing agent used in the cotton textile industry, Emsize E1 (Emsland-Starke GmbH, Germany), were used as carbon source. The latter's stock solution (100 g L⁻¹) was prepared by hydrolyzing Emsize E1 in alkaline conditions (Lourenço et al. 2000). The azo dye stock solution (5.0 g L⁻¹) was prepared by dissolving Acid Red 14 (AR14, Chromotrope FB, Sigma-Aldrich, 50% dye content) in distilled water.

The synthetic textile wastewater used as feed solution was prepared by diluting the carbon source stock solution with a nutrients and salts solution, to a COD content of 1,000 mg O₂ L⁻¹ (1.15 g L⁻¹ Emsize E1). When acetate was used as carbon source, it was directly dissolved in this solution, to this same overall COD level. This solution contained pH buffering phosphates and nutrients to the following concentrations: 2,310 mg L⁻¹ Na₂HPO₄.12H₂O, 762 mg L⁻¹ KH₂PO₄, 143 mg L⁻¹ NH₄Cl, 22.5 mg L⁻¹ MgSO₄.7H₂O, 27.5 mg L⁻¹ CaCl₂, 250 μg L⁻¹ FeCl₃.6H₂O, 40 μg L⁻¹ MnSO₄.4H₂O, 57 μg L⁻¹ H₃BO₃, 43 μg L⁻¹ ZnSO₄.7H₂O, 35 μg L⁻¹ (NH₄)₆Mo₇O₂₄.4H₂O. All salts were analytical grade. In this feed solution the COD:N:P mass ratio was 100:3.7:37. Dye was added from the stock solution to a concentration of 40 mg AR14 L⁻¹.

The feed was supplied to the SBRs with a hydraulic retention time of 12 h. Three-fold increases in dye and Emsize (the latter together with NH₄Cl) concentrations in the SBR feed, to 120 mg AR14 L⁻¹ and to an organic loading rate (OLR) of 6.0 kg COD m⁻³ d⁻¹, were applied on days 158–164 (period B) and 165–172 (period C), respectively (Table 1). Following a transitional period, when the original operating conditions were restored (days 173–179, period A), acetate was introduced in the feed, the carbon source being changed from 100% to 50% Emsize (days 256–291, period E), and finally to 100% acetate (days 292–314, period F). In fact, starches (such as Emsize E1) and acetic acid (acetate as conjugate base) are among the major pollutants found in textile wastewater contributing for the high organic load characteristic of these effluents, arising from desizing, scouring, washing and dyeing processes (Delée et al. 1998).

Total suspended solids (TSS), volatile suspended solids (VSS), sludge volume index after 5 and 10 min settling (SVI₅ and SVI₃₀, respectively), COD and pH were determined through standard procedures (APHA 1995). Biomass morphological analysis employed a transmission light microscope (BA200, Motic) with a digital camera and respective software (Moticam 2, Motic). Staining of intracellular lipids was performed with Nile Blue according to Ostle & Holt (1982), visualized through fluorescence microscopy (BA410 microscope with episcopic fluorescence attachment EF-UPR-III and TRITC (Rhodamine)/Dil/Cy3 filter set, Motic).

Color removal was followed by reading the absorbance of clarified (centrifuged) samples at 515 nm (AR14 maximal absorbance wavelength in the visible region) in a UV-visible spectrophotometer (Specord 200, Analytik Jena) against distilled water. Dye concentration in terms of color equivalents (dye-color) was determined from a calibration curve obtained with standard AR14 solutions. Dye degradation and metabolite formation were followed by reversed-phase high performance liquid chromatography (HPLC) (Franca et al. 2015).

Inoculation with a 2.5-month stored AGS led to different biomass profiles in the two SBRs, with distinct settling characteristics. SBR1 presented a peak in the SVI₅ and SVI₃₀ values on day 48 (177 and 98 mL gTSS⁻¹, respectively), subsequently stabilizing along period A at higher levels than those registered in SBR2 (Table 2). Although SVI was more stable in the latter during period A, the shock loads applied in periods B–C caused a sharp increase in the SBR2 SVI values, high levels being maintained throughout the subsequent period D (Table 2). As a result of this settleability shift, TSS levels decreased from 7 to 5 gTSS L⁻¹ in SBR2 after period C, in contrast with SBR1, which maintained an average value of 5 gTSS L⁻¹ throughout periods A–D (Table 2). Accordingly, while the sludge retention time (SRT) levels in SBR1 were mostly stable until the end of period D (9 days, on average), in SBR2, the initial average sludge age of 18 days deteriorated to 6 days after the shock loads were applied (Table 2). Subsequently, upon introduction of acetate as 50% and 100% of the carbon source load in the feed (periods E and F, respectively), biomass settleability was significantly reduced in both reactors (Table 2), reaching maximal SVI₅ and SVI₃₀ values of 592 and 282 mL gTSS⁻¹ in SBR1, and 400 and 255 mL gTSS⁻¹ in SBR2, respectively, on day 299. This substrate switch led to an almost complete biomass washout, resulting in biomass concentrations around 1.5 gTSS L⁻¹ in both SBRs. Yet, after adaptation to the new carbon source, SBR2 resumed biomass accumulation along period F, with SVI₅ and SVI₃₀ levels gradually decreasing to final values of 225 and 123 mL gTSS L⁻¹, respectively.

Overall, the average and standard deviation values for the TSS and SVI in each experimental period (Table 2) showed that for both SBR1 and SBR2, periods A, B and C presented similar results in terms of biomass properties. In contrast, significant differences were observed in periods D, E and F for the two reactors.

The food-to-microorganism ratio (F/M) values in SBR1 and SBR2 (Table 1) follow the same tendency along periods A–D. In periods A and B, the F/M values were within the range of 0.3–0.5 gCOD gVSS⁻¹ d⁻¹. Higher values were achieved in period C (1.2 and 0.9 gCOD gVSS⁻¹ d⁻¹ in SBR1 and SBR2, respectively), when a 3-fold higher organic load was applied, subsequently decreasing upon reestablishment of the initial feed conditions (period D: 0.4 and 0.5 gCOD gVSS⁻¹ d⁻¹ in SBR1 and SBR2, respectively). Accordingly, the biomass morphology evolved similarly in the two reactors along these periods (Figure 2).

Upon introduction of acetate as carbon source (periods E and F), significant differences were observed in the biomass morphology in the two reactors (Figure 2). In fact, the F/M increased (Table 1) more significantly in SBR1 (periods E and F: 1.1 and 1.8 gCOD gVSS⁻¹ d⁻¹, respectively) than in SBR2 (periods E and F: 0.5 and 1.0 gCOD gVSS⁻¹ d⁻¹, respectively). Similar to the apparent lower density of the granules microscopically observed, these F/M values denote the deterioration in the sludge settleability, especially in SBR1.

The low average SRT levels observed (Table 2) were mainly caused by erratic, intense biomass washout episodes in the SBRs. This likely indicated the dominance of heterotrophic fast-growing organisms, which have been associated with AG instability (Pronk et al. 2015a). High microbial growth rates are a consequence of high residual COD levels at the start of the aeration phase (de Kreuk & van Loosdrecht 2004). Filamentous bacteria are usually associated to these operational conditions, but none were observed during this study. This might result from the starvation stage at the end of the aerobic phase, since the feast–famine regimen has been shown to prevent filamentous outgrowth in AG (de Kreuk & van Loosdrecht 2004). In addition, the shear stress caused by mechanical stirring might also have impaired the development of filamentous organisms.

The change to 50% acetate in the feed did not affect decolorization, and even improved overall COD removal, mostly aerobically (Figure 3(e)). Upon complete replacement of Emsize with acetate, color removal dropped to 16% in SBR1 (days 299–307), being subsequently restored to regular levels from day 309 on. In contrast, color removal in SBR2 responded positively to acetate, namely during the plug-flow feeding period, reaching an overall yield of 91% (Figure 3(f)). In fact, the color profile representative of period F suggests that the anaerobic plug-flow feeding through the sludge bed can be advantageous in terms of decolorization performance when acetate is present in the wastewater as the main carbon source. In addition, the introduction of acetate caused a pH shift to higher values (Figure 3(e) and 3(f)). In fact, the pH level was stable throughout the anaerobic phase, denoting the presence of a non-fermentable substrate, the latter being rapidly consumed upon the start of aeration, causing an equally fast pH increase.

The combination of substrate diffusion and microbial consumption rates is known to determine the substrate gradient inside a biofilm and, consequently, its stability (de Kreuk et al. 2010). Specifically, a plug-flow hydrodynamic regimen rising through the settled bed, with high substrate concentration allows soluble, easily biodegradable substrates to penetrate the entire AG depth and be converted to storage polymers (Pronk et al. 2015a). Consequently, easily biodegradable COD will not be available in the subsequent SBR aerobic phase and slow microbial growth can occur down to the core of the AG, ensuring a stable AG structure. In contrast, in an anaerobic fast feeding-stirred strategy the biomass contacts with a lower substrate concentration, as a result of its dilution with the remaining supernatant from the previous SBR cycle. Thus, in this situation, substrate gradients are likely to occur in the AG, which may lead to AG instability (de Kreuk et al. 2010).

In addition to the feeding regimen, the reactor geometric configuration should be taken into account. In fact, Rocktäschel et al. (2013) noted that in reactors with a small H/D, the plug-flow feeding may develop heterogeneous flow patterns through the granular bed, leading to short contact times between the substrate and the biomass and thus to incomplete anaerobic carbon uptake during this stage. Accordingly, low COD removal occurred during the plug-flow anaerobic feeding of SBR2, with most of the anaerobic COD removal being achieved during the subsequent stirred anaerobic phase, where better mass transfer conditions were provided. Owing to a lower microbial growth rate in the aerobic phase, complete COD removal under anaerobic conditions has been shown to increase AG density and structural stability (Pronk et al. 2015a). In fact, despite the negligible COD removal during the plug-flow feeding phase of SBR2, the subsequent stirred anaerobic phase led to faster anaerobic COD removal than in SBR1. This might explain the lower average SVI values in SBR2, from period D on (Table 2). Yet, complete anaerobic COD removal was never achieved in the SBRs, and this situation worsened when the carbon source was changed from hydrolyzed hydroxypropyl starch (fermentable substrate) to acetate (VFA).

The effects of the feeding strategy and feed composition on the stability of AGS in the treatment of textile wastewater were assessed in two non-tubular SBRs, a statically fed, anaerobic–aerobic SBR (SBR1), and an anaerobic plug-flow fed, anaerobic–aerobic SBR (SBR2), the latter allowing the biomass to contact with the feed during the fill stage. Overall, high color and COD removal levels were attained in both SBRs. However, dye and COD shock loads had a negative effect on AG integrity and further aerobic bioconversion of the amine metabolite was not observed. COD uptake was minimal in the plug-flow feeding stage of SBR2, but the fraction of COD removed anaerobically was generally higher than in SBR1. The change from the hydrolyzed starch-based to the acetate-based feed resulted in AGS settleability deterioration in both SBRs. After adaptation, lower SVIs were attained in SBR2 and decolorization yields during its plug-flow feeding period improved, revealing a higher capacity of SBR2 to deal with substrate-related variations.

This work was financed by Fundação para a Ciência e a Tecnologia (FCT, Portugal), projects PTDC/EBB-EBI/120624/2010, PTDC/AAG-TEC/4501/2014 and UID/BIO/04565/2013. Funding from Programa Operacional Regional de Lisboa 2020 (Project No. 007317) is also acknowledged. R. D. G. Franca and N. D. Lourenço acknowledge FCT for doctoral (SFRH/BD/95415/2013) and post-doctoral (SFRH/BPD/88095/2012) research grants, respectively.