The influence of excess sludge discharge on the performance of a full-scale UASB reactor

Anaerobic digestion (AD) was previously considered to treat only high-strength wastewaters at temperatures above 20–25 °C (Bergamo et al. 2009). Climatic conditions in countries like Brazil are favorable for AD and this technology has been used to treat low-strength wastewaters, such as domestic sewage, since the 1980s. AD became more widespread with the introduction of upflow anaerobic sludge blanket (UASB) reactors (Lettinga et al. 1980). At present, anaerobic treatment of wastewaters is widely accepted as reliable and is used extensively in practice.

However, it is known that compared to an aerobic system, anaerobic processes have slow growth rates, mainly associated with methanogenic archaea. So, long solids retention times are required because only a small portion of the degradable organic waste is synthesized into new cells. Inside reactors, a dense sludge bed is established at the bottom, and the majority of the biological transformations in it are effected by bacteria in aggregates of flocs and granules. These aggregates have good settling properties and are not susceptible to system washout for practical reactor conditions (Veeresh et al. 2005; Latif et al. 2011). UASB reactors also play an important role in wastewater treatment in tropical countries because of the favorable temperature conditions (Kato et al. 2003; von Sperling et al. 2005; Chernicharo 2006).

Since about 1985, analytical and computational developments have increased understanding of the parameters and operating conditions that influence anaerobic reactor performance. Nevertheless, several parameters, such as the cellular yield coefficient (Y), need further investigation to improve AD efficiency. Factors such as the composition of the wastewater, the temperature, and sludge age all influence performance (Chang & Lin 2004; Tawfik et al. 2010), resulting in a wide range for Y. Values of 0.05–0.25 have been used in design and modeling (Gavala et al. 2003). Few studies consider controlling sludge age by measuring the discharge of excess sludge produced during the treatment of domestic sewage under mesophilic conditions in full-scale UASB reactors. The resulting wide range of Y values can lead to under- or over-sizing sludge storage devices, and can directly affect monitoring and full-scale operational activities.

Sludge blanket depth plays an important role in the design and performance of UASB reactors. The importance of sludge discharge is linked to the production of biomass, as well as foam layer formation and solids loss in the effluent. When excess sludge is not periodically discharged, the sludge bed increases and the fluidized zone decreases. As a result, sludge that is washed out with the effluent compromises the quality of the treated effluent (Kalogo & Verstraete 1999). Discharging the sludge serves two functions: ensuring the quality of the treated effluent and promoting cell renewal. Thus, the frequency of sludge discharge represents an important operating parameter for UASB reactors.

Few studies investigate the influence of sludge discharge frequency on operational performance and cellular Y in full-scale reactors; the generally established design sludge discharge frequency has not been verified in practice. Furthermore, the operational conditions are variable, and distinct from those of the design or theoretical parameters. The objective of this study was to evaluate the influence of sludge discharge frequency on the performance and Y values of a UASB reactor operating in a full-scale domestic sewage treatment process under mesophilic conditions.

This study was conducted at the Mangueira Wastewater Treatment Plant (MWTP) (Figure 1), which is in a low-income region of Recife, Northeast Brazil. The MWTP was designed to serve a population of 18,000 inhabitants and consisted of a grit chamber, a UASB reactor (total volume 810 m³) with an 8 hour hydraulic retention time (HRT) and average flow of 32 L s⁻¹, and a polishing pond (d = 1.5 m and HRT = 3.5 days). The reactor contained eight compartments, each with a volume of 101.25 m³ and a useful height of 5 m. Each compartment was operated as an individual reactor, and the eight compartments operated in parallel to allow for operational flexibility and variation. Eight corresponding drying beds were used to dewater the sludge from the compartments (Florencio et al. 2001).

The influent to the treatment plant was pumped into an elevated chamber that contained a grid, and gravitational flow directed the wastewater on to the subsequent units. After the grit chamber and Parshall flume, the influent was distributed to a central box, then two intermediary distribution boxes, and finally eight small boxes feeding the individual reactor compartments from nine points. Each compartment collected its treated effluent through perforated tubes in the upper part and discharged it to a lateral canal outside the reactor. After the effluents were merged, the water was polished in a pond for post-treatment.

Sludge sampling and disposal were performed using five pipes placed at different levels within the reactors (0.3, 0.8, 1.3, 1.8, and 2.3 m, from the bottom). Each drying bed had a surface area of 25.2 m², in which excess water in the sludge was evaporated off. Thus, the amount of sludge removed from the compartments was limited to the drying bed capacity, approximately 9% of the total useful volume of a compartment.

Before initiating the study, the sludge was simultaneously discharged from all compartments to start the cycles without excess accumulated sludge. The discharge cycles then commenced, and the period of each compartment was set at between 1 and 8 months according to the compartment number; i.e., sludge was discharged from compartment 1 on a monthly basis (30 days), compartment 2 every 2 months (60 days), and so on until compartment 8, which was discharged every 8 months (240 days).

At the end of each established period, the sludge was discharged by sequentially opening the highest to lowest compartment ports. After reaching the capacity of the drying bed (8.82 m³), the pipe ports were closed in the same order and the discharge operation was complete. Before sludge discharge began, a profile of the volatile suspended solids (VSS) was obtained by collecting sludge samples from the five pipes, in order to assess the total amount of biomass in each compartment prior to discharge.

Reactor performance was assessed using – chemical oxygen demand (COD) (total and filtered through a 0.45-μm fiberglass membrane), VSS, and pH. These parameters were monitored using raw sewage collected after the grit chamber and in the effluent of each reactor compartment, using the APHA standard procedures (2005).

Effluent quality in the compartments was determined on the basis of COD removal efficiency, calculated as the difference between the total influent COD and filtered effluent COD (Lucena et al. 2011). The influent and effluent monitoring of COD and VSS consisted of sampling at least twice weekly regardless of the compartment concerned. For the organic loading and VSS mass calculations, the average reactor influent flow for a selected period was calculated using measurements that were taken at least three times each week. The effluent flow from each compartment was determined from the total flow divided by eight.

where MX_i, biomass at the beginning of the period after sludge discharge in the previous period (kg VSS); MX_f, biomass at the end of the period before sludge discharge in the next period (kg VSS); MX_eff, biomass lost in the effluent during the period (kg VSS); MX_fpp, biomass at the end of the previous period (kg VSS); MX_dpp, biomass discharged at the end of the previous period (kg VSS).

The total sludge mass contained in each compartment at the end of any period was defined as the sum of the product of the VSS concentration at each of the five levels obtained from the sludge profiles and the volume of the corresponding sludge sub-layer.

The minimum sludge mass – in kg VSS – i.e., the amount that must be contained in the compartments to prevent operational instability in the system, was calculated by dividing the organic load that was removed daily (L_o, kg COD_removed d⁻¹) by the specific methanogenic activity (SMA, kg COD kg VSS d⁻¹) obtained previously with the same sludge and reactor (Florencio et al. 2001; Kato et al. 2003; Lucena et al. 2011). The number of discharges required to maintain the minimum sludge mass was studied based on the quantity of sludge mass contained in the compartments before discharge and after removal of 9% of the reactor volume and the amount of biomass produced in the interval between successive discharges. Thus, the ideal interval between discharges was determined using the ratio of the quantity of sludge discharged to the daily mass of sludge accumulated in the system.

The COD removal efficiency and VSS concentrations in the effluent of each compartment are shown in Figure 3. Compartment 1 (Figure 3(a)) exhibited the highest operational stability in terms of both COD removal efficiency and VSS content in the effluent. This confirms that the higher frequency of sludge discharge results in more frequent renewal of bacteria and removal of inert materials from the compartment. It also demonstrates that the amount of sludge removed from the reactor minimized variations in COD removal efficiency and effluent VSS values, regardless of discharge frequency. However, soon after the periodic discharges from them, compartments 2 to 8 (Figure 3(b)–3(h)) exhibited decreased COD removal efficiencies and increased variations in effluent VSS levels. This is probably because of influent variations (flow rate and organic load) and pumping interruptions that occurred at times during the study. These resulted in a high standard deviation (SD ±189 mg L⁻¹) for the total COD influent (n = 520). However, the effects of operational problems can be minimized by renewing the compartment sludge more frequently, as demonstrated by the performance of compartment 1 (Figure 3(a)). Nevertheless, the majority of the COD removal efficiencies remained above 60% in all compartments.

Mahmoud et al. (2004) also reported that highly variable COD removal efficiencies were influenced by the considerable fluctuations in the sewage concentration and, thus, in sludge loading rates. Such fluctuations occur more often during rainy periods, and also caused changes in COD composition, which caused a significant proportion of colloidal particles to be retained in the sludge bed and, thus, temporary negative values in efficiency. Halalsheh et al. (2005) assessed the treatment of domestic sewage in UASB reactors and found that regular discharges of 25–30% of the total sludge did not improve the removal of solids or the performance of the UASB reactor. Their results did not indicate any significant increase in COD removal efficiency; and regular discharges were not as beneficial for system performance as expected. Bhunia & Ghangrekar (2008) obtained random variations in COD removal efficiency and observed sludge washout in all experimental runs during the first 2–3 weeks of operation. However, COD removal and sludge washout were stabilized after reaching steady-state conditions.

The best effluent VSS concentration was observed for compartment 1 (Figure 3(a)), where fluctuations were lower than in other compartments. During the study, effluent VSS concentrations decreased slightly soon after a discharge, for all compartments, but increased subsequently until the next discharge.

pH values were nearly neutral (approximately 6.9) in all compartments and discharge frequency had no significant influence. This neutral pH indicated that the microbial populations were balanced and presented no significant risk of acidification.

Figure 4 presents the values for the cellular Y calculated as a function of the discharge frequency in months (Equation 1). In general, lower Y values were observed when higher sludge discharge frequencies were used – e.g., in compartments 1 and 2 (Figure 4(a) and 4(b), respectively). The Y values obtained for compartments 1 and 2 were between 0.03 and 0.18 kg VSS kg COD_removed⁻¹. Most results from compartment 3 are also within this range, which was also found by Tawfik et al. (2010) for UASB reactors treating domestic sewage (0.06 kg VSS kg COD_removed⁻¹) and Chang et al. (2004) for UASB reactors used for hydrogen production from sucrose (0.10 kg VSS kg COD_removed⁻¹).

Figure 4(d)–4(h) demonstrate that the discharge frequencies applied to compartments 4–8 inclusive had a greater influence on biomass production because cell yields were higher than in compartments 1–3. The 8-month discharge frequency in compartment 8 resulted in the highest value of Y (0.4 kg VSS kg COD_removed⁻¹), which is near that typically observed in conventional activated sludge systems (0.5 kg VSS kg COD_removed⁻¹) (Wei et al. 2003).

Figure 5 presents the variation in the average values of cellular Y (Figure 5(a)) and VSS concentration in the effluent (Figure 5(b)) as a function of the time interval between successive discharges. Higher mean values of Y produce higher VSS losses in the effluent. The values varied from 25 to 125 mg VSS L⁻¹ in the effluents from compartments 1 to 8, respectively. The highest values can be explained by sludge accumulation during the longer periods; a greater sludge volume results in a bed with excess sludge that can be washed out more easily because of turbulence from gas and liquid upflow. Therefore, low levels of effluent VSS in UASB reactors can be ensured by high sludge discharge frequencies.

The average VSS concentration of the sludge retained in the reactor compartments (Figure 6) was 18–22 kg VSS m⁻³ for discharge periods of up to 5 months and 25–29 kg VSS m⁻³ for longer intervals. These are relatively low compared with those of other full-scale UASB reactors treating high-strength wastewaters. This can be explained by the low total COD in the influent (average 303 mg L⁻¹; SD ± 84.75 mg L⁻¹; n = 520) to the MWTP resulting in low organic loading rates. Nevertheless, the values found in the present study agree with the concept that biomass production results in higher sludge accumulations when the discharge frequency is lower.

Although the volume of excess sludge discharged was limited to approximately 9% of the total useful volume of each compartment, the minimum mass in the compartments was always met (Table 1) to maintain operational stability. The results indicate that treatment stability improves (as measured by COD removal efficiency and VSS concentrations in the effluent, Figures 3(a) and 5(b), respectively) with increases in sludge discharge frequency. Thus, the number of discharges required to reach the minimum mass of sludge inside the reactor that would guarantee operational stability was estimated in kg VSS. As previously described, the discharge number was calculated based on the volume of sludge removed in each discharge and the amount of biomass produced during the interval between successive discharges. The number of discharges required to reduce the volume of sludge to the required minimum was not reached in any compartment, including compartment 1, which was subject to 17 discharges over the course of the experiment. Therefore, a higher volume of excess sludge must be discharged during each discharge applied or the number of discharges must be increased.

A lower period or higher frequency of excess sludge discharge minimized variations in COD removal and the concentration of volatile solids in the effluent. With periodic sludge discharges, the removal efficiency of organic matter (as COD) remained within operational standards (>60%) observed in practice. Although satisfactory COD removal efficiency was achieved for all compartments, the sludge discharge frequency influenced their operational stability. The compartment with monthly discharges exhibited the greatest stability in terms of COD removal (efficiency approximately 80%) and effluent VSS concentration (25 mg L⁻¹). Nevertheless, in the case of domestic sewage with similar influent concentrations, periods between sludge discharges up to four months can be applied without significant loss of performance stability. The lowest cellular Y was observed when higher sludge discharge frequencies were applied (Y between 0.03 and 0.14 kg VSS kg COD_removed⁻¹ for monthly discharge). The overall results confirm that, in practice, controlling the discharge frequency is important in maintaining an adequate amount of active biomass inside a UASB reactor and to preserve effluent VSS concentrations.

The authors thank the Pernambuco Sanitation Company (COMPESA) and the Brazilian agencies CNPq (CT-HIDRO program), FACEPE, and FINEP for their financial support to the Laboratory of Environmental Sanitation of the Federal University of Pernambuco (LSA-UFPE), as well as the kind help of colleagues in the LSA-UFPE for their field and analytical efforts, and assistance in drafting figures and processing data.