Gravimetric selection of activated sludge for settling properties improvement and granular sludge formation – full scale case study

One of the main hydraulic bottlenecks of activated sludge (AS) systems is efficiency of solid-liquid separation in secondary settling tanks (SSTs). In the event of wet weather inflow, SSTs with poorly settling sludge are prone to sludge blanket rise potentially leading to biomass washout. The number of factors influencing sludge settleability and compressibility is high (Jin et al. 2003) and for BNR systems with long sludge retention time (SRT), filamentous bulking is a frequent challenge (Gabb et al. 1991; Jenkins et al. 2004). Since some factors promoting filamentous outgrowth (interpreted as filamentous bacteria proliferation to the extent of interfering with sludge settling and compaction), such as temperature drop (Knoop & Kunst 1998), are beyond operators’ control, novel solutions that help improve AS settling characteristics are in demand. One of most promising technologies for improving sludge sedimentation rate is aerobic granular biomass (AGB) (Nor Anuar et al. 2007). AGB has successfully been cultivated in full-scale sequencing batch reactor (SBR) systems (Pronk et al. 2015) using hydrodynamic shear force and feast-famine feeding regimes (Nancharaiah & Reddy 2018). Currently efforts are being made to achieve improve sludge settling characteristics in continuous flow systems by implementing two-zone sedimentation tank for selective retaining fast settling agglomerates and adding micropowder to induce bacterial attachment (Zou et al. 2018), introducing settling velocity selector and feed/famine conditions (Sun et al. 2019), seeding reactors with AGB and applying size-based selection (Corsino et al. 2016) or metabolic selection pressure while cultivating phosphate accumulation organisms (Devlin & Oleszkiewicz 2018). Successful laboratory scale continuous flow AGB systems required more than one granulation driver with settling velocity or particle size being the primary one. This led to the concept of using hydrocyclones – an external gravimetric selector previously used for full-scale deamonnification reactors (Klein et al. 2012; Wett et al. 2013) to selectively retain larger, denser particles.

The hydrocyclone is a simple mechanical device without moving parts that separates solid mixtures based on size and specific gravity. Thanks to its cono-cylindrical shape and tangential feed inlet it transforms feed pressure into centrifugal force (Bradley 2013). Particles entering a device spin in a high-velocity vortex and are separated based on terminal settling velocity. When fed with AS, a mixture of non-uniform floc densities and sizes, the hydrocyclone separates the feed into two fractions: underflow containing denser and/or larger flocs and overflow with lighter and/or smaller flocs. Hydrocyclone implementation in AS systems was the object of a few laboratory-scale and pilot research projects with a focus on sludge disintegration and releasing internal carbon sources (Liu et al. 2017; Xu et al. 2018, 2019). Xu et al. (2019) reported a slight improvement in sludge settling characteristics but that no AGB formation occurred while mean floc diameter decreased. Results of the successful full-scale hydrocyclone implementation in a AS system with low SRT was reported by Avila et al. (2021) were both biological and physical selection pressures were applied.

The objective of this study is to investigate potential for gravimetric selection technology conducted in a full-scale hydrocyclone installation to improve settling characteristic and achieve full or partial AGB development in a high-SRT BNR system, without changing biological selective pressure. Additionally, the hydrocyclone's ability to selectively remove filamentous bacteria by means of shear force was explored.

The system under study is an A2O AS WWTP with enhanced biological nutrient removal coupled with chemical phosphorus precipitation It treats on average 140,000 m³/d with a load of 1,050,000 P.E. A mean SRT of 25.9 ± 4.1 days was maintained. Biological treatment consists of five process lines (main operational parameters are presented in Table 1) with three independent sludge recirculation systems. Two process lines – experimental train – were equipped with gravimetric selection installation with an overall capacity of 5,400 m³/d placed in the return activated sludge (RAS) line. One process line – identical to experimental train – is referred to as the ‘Reference train’ and the other two due to difference in reactor configuration were not used in this study. The flow diagram of the experimental process train and hydrocyclone dimensions are presented in Figure 1. Approximately 1,400 m³/d was pumped through the hydrocyclone installation and separated into two fractions: retained underflow (UF) and overflow (OF), which is removed as waste activated sludge (WAS). Daily amount of WAS from each process line was set by process engineer. Installation was fed sequentially following periodical WAS removal. When fed, constant inflow of 155 m³/h and stable UF/Feed split ratio of 0.21 ± 0.04 in the hydrocyclone was maintained throughout the whole operation period under discussion.

The quality of the effluent 24 h-composite samples in the experimental process train and plant effluent was measured with standard photometric cuvette tests (Hach Lange GmbH) and DR6000 spectrophotometer (Hach Lange GmbH): total nitrogen (LCK 338), total phosphorus (LCK 350), chemical oxygen demand (LCK 114). The concentration of total suspended solids (TSS) was measured according to Standard Methods. The sludge volume index (SVI) after 30 minutes of settling (SVI₃₀) value was monitored according to APHA (2012) and used as a simple indicator of sludge settling characteristic throughout the whole trial period. The state of granulation was described by comparison of SVI after 5 minutes of settling (dSVI5) and 30 (dSVI30) minutes of settling sludge sample diluted to approximately 2 g/L (Etterer & Wilderer 2001). Floc size distribution was monitored using automated image analysis with a Malvern Morphologi G3 analyzer. Microscopic examination of the AS samples were performed according to Eikelboom (2000) and filamentous index (FI) was used as a parameter for quantification of filamentous microorganism population in the sludge.

Long-term operation of the hydrocyclone installation did not impact overall treatment efficiency. As presented in Figure 2, effluent quality – expressed as weekly averages – in the experimental process train did not deviate from plant effluent. The experimental process train effluent constitutes 40% of total plant capacity but since there is no significant, long-term difference between the two data sets in Figure 1 we assume that treatment process in both experimental and reference process train had comparable efficiency.

Hydrocyclone installation start-up began in October 2019 and the first 60 weeks of operation data was analysed. The difference between SVI₃₀ variability in experimental vs. reference process train is presented in Figure 3. SVI₃₀ of the reference plant shows an increase when wastewater temperature drops below 19 °C, which is typical for this plant, and a decrease from a value of 190 mL/g to 120 mL/g between week 21 and 22 as a result of aluminium sulfate dosing. Gravimetric selection technology significantly reduced experimental SVI₃₀ values. The first quantifiable impact on performance was visible when a period equal to two SRTs or approximately 7 weeks had passed. The SVI₃₀ dropped noticeably in week 8 and then again in week 35. Settling characteristics were not stable and a sharp rise in SVI started in week 10 and 52. It is hypothesised that the reason behind this is seasonal filamentous outgrowth which happened concurrently in the reference process train. Noteworthy, despite dropping process temperature, installation start-up allowed for temporary improvement of settling properties. Improvement was not permanent and when SVI values rose to a similar SVI level as that of the reference train, the rise was of a much larger relative magnitude. The reason for the temporary improvement in SVI values during lower process temperature is a subject for further research since during the next cold season the SVI values rose and were stable during the whole season. (data not shown). In the experimental system, the poor settling period lasted till week 34 when a second rapid decrease in SVI₃₀ took place. SVI₃₀ values in the experimental process train dropped below those in the reference train, reached the level of 50 ml/g and remained stable during next 10 weeks. Based on the presented results, the beneficial, but seasonal effect of gravimetric selection technology on AS settling properties is evident. Between week 36 and 52 (warm season) the average SVI₃₀ values in in experimental and referenced AS were 61 mL/g and 115 mL/g, respectively, so that the value in the experimental system was 47% lower. Avila et al. (2021) reported 45% reduction in SVI in a system with much lower SRT, where AS selection pressure exerted by hydrocyclone is much higher. The data presented here show that a full-scale hydrocyclone installation in a high-SRT system can seasonally improve sludge settling characteristics characteristics to the same extent.

Table 2 presents FI values in sludge samples from experimental and reference process train. Since gravimetric selection conducted by hydrocyclones could potentially selectively remove filamentous microorganisms, additional microscopic analysis was done in WAS from experimental sludge samples. Although an average difference of 0.5 between experimental and reference AS can be noted, FI values for WAS removed from the system did not deviate from those in the AS, suggesting that the hydrocyclone did not physically separate filaments from the AS. It is worth noting that differences in FI between experimental and reference train should not be the sole indicator of different settling properties. As can be seen in Figure 3 and Table 2 when SVI in experimental train deviated from reference one by week 11, FI values in all three samples were the same. Other factors such as flocs size and density play a major role in sludge settling characteristics, but would not be captured by FI From the results given it can be concluded that filamentous bacteria cannot, in this system setup and operating conditions, be selectively removed by a hydrocyclone. It is hypothesised that filamentous microorganisms in AS has a twofold effect when improvement of settling properties by means of gravimetric selection is considered. Low levels of filaments in activated sludge is believed to act as a backbone to the floc structure. As stated by Burger et al. (2017) filament abundance increases the floc strength factor and therefore its resistance to shear stress. This would suggest that flocs with a certain filament content may be less susceptible to shear and centrifugal forces in hydrocyclone and therefore are more likely to form larger, stable flocs that might turn into granules. On the other hand, since hydrocyclone could not selectively remove filaments, this technology did not control filamentous outgrowth and its destructive effect on sludge settling properties when they start to protrude from the floc (Wágner et al. 2015).

As typical AS flocs switch to granule form SVI₅ values decrease and approach SVI₃₀ values. Figure 4 presents dSVI₅ and dSVI₃₀ data in experimental train. It could be seen that although during the whole analysed period that dSVI₅ values were continuously higher than dSVI₃₀ (80 ml/g between 18 and 22 week)_, after 36 weeks' difference between them was reduced to 30 ml/g.

As seen in Figure 5 floc size distribution shifted. Floc median diameter increased from 200 μm in week 20 to 400 μm in week 40 when SVI₃₀ dropped to 50 ml/g. Aggregates larger than 500 μm constituted only 5% of all flocs volume in week 20 but increased to 30 and 25% in week 40 and 60, respectively. In addition, a change in sludge morphology is confirmed by the more frequent occurrence of granule-like aggregates with diameter exceeding 1,000 μm. Sludge granulation was not complete: in week 40, with the lowest SVI₃₀ values, 50% of all flocs by volume remained smaller than 400 μm. It is critical to note that from an operator's point of view full granulation in a continuous flow system might not be desirable. Absence of smaller flocs in SST, that form continuous matrix during settling, trapping dispersed biomass, might result in elevated effluent TSS concentrations (Pronk et al. 2015; Avila et al. 2021).

Results of implementing gravimetric selection of recycled activated sludge presented in this full-scale case study as well as that by Avila et al. (2021) contradict findings of a lab-scale experiment from Xu et al. (2019) were hydrocyclone operation decreased average floc size and no granules occurred. Reasons behind that might be different time-scale, operation condition such as overflow/feed ratio or lack of mineral precipitates that in full scale demonstration might have acts as granule seed. All of the above prove that exact mechanism behind hydrocyclone AS selection is not yet fully understood and its implementation in different wastewater treatment systems might have unknown outcomes.

A full-scale implementation of activated sludge gravimetric selection technology was demonstrated at large BNR WWTP operated at high SRT with an aim to improve SST performance by changing sludge morphology. A hydrocyclones installation has been in operation since October 2019 and 60 weeks of operational data from both experimental and reference process trains were analysed. The evaluation of the effects of the technology on activated sludge morphology, settling properties and an overall plant operation lead to the following conclusions:

• The start-up phase lasted app. two SRTs when after the initial phase quick improvement in settling characteristics occurred with an average decrease in SVI of 14 ml/g every week.

• Shear forces acting in hydrocyclones did not prevent filamentous growth during the cold season.

• Aside from the cold season, gravimetric selection technology proved to be able to significantly improve settling characteristics of AS and increase the hydraulic capacity of the SST and therefore assure safe wet-weather operation.

• Long-term hydrocyclone operation resulted in a clear change in sludge morphology. Sludge in the experimental train had a significantly higher frequency in granule occurrence.

All relevant data are included in the paper or its Supplementary Information.