Start-up of a decentralized pilot plant for the anaerobic treatment of domestic wastewater

Effective household water cycles are important for reducing water use and providing potable water. Treating and recycling domestic wastewater for reuse for non-potable domestic purposes like flushing toilets or washing cloths is a promising approach. Up to 60% of household potable supplies can be replaced by non-potable water (Herbst 2008). In this context decentralized systems are required that treat wastewater on site, avoiding both the need to transport water long distances and the related maintenance costs. Decentralized plants should be designed to be space efficient and odourless, as well as easy to use, so that simple operating instructions are sufficient (Herbst 2008; Massoud et al. 2009).

Due to low energy consumption, low sludge production, low space requirements and the production of useable energy as biogas, anaerobic digestion is particularly attractive for organic removal in decentralized wastewater treatment systems (Olsson et al. 2005; Batstone 2006). Depending on whether the process design comprises one- or two- stage operation, and the operating conditions, including temperature and pH, and the reactor types, organic removal efficiency in anaerobic digestion can be between 25 and 90%. Kobayashi et al. (1983), for example, achieved 73% organic removal efficiency for low strength domestic wastewater (288 mg-COD/l) using a lab-scale aerobic filter with a capacity of 11 ml/min. With a 120 liter upflow anaerobic sludge blanket reactor, Barbosa & Sant'Anna (1989) achieved 74% average organic removal treating raw domestic sewage (627 mg-COD/l) with 4 hr HRT. Donoso-Bravo et al. (2009) treated low-strength wastewater (500 mg-COD/l) with a high proportion of particulate organic matter (70%, COD basis) using a two-stage system consisting of anaerobic sequencing batch reactors. 69 and 50% COD removal efficiencies were achieved for organic loading rates of 630 and 1,220 mg-COD/l.

Carbamide, present in domestic wastewater due to urine, requires an additional treatment step after anaerobic digestion to reduce the ammonia concentration. Anaerobic ammonium oxidation (Anammox) is an innovative possibility for ammonia removal, in which it is converted to elementary nitrogen anaerobically by using nitrite (van de Graaf et al. 1996; Ma et al. 2016). Fux et al. (2002) treated wastewater containing about 650 mg-NH₄-N/l in a sequencing batch reactor with a capacity of 2 m³. Nitrogen removal efficiencies of 85 to 99% and reduction rates between 1,100 and 1,200 mg-NH₄-N/l/d were achieved. Van der Star et al. (2007) successfully started a 70 m³ reactor treating 10,715 mg-NH₄-N/l/d using 5 m³ of biomass from an enrichment reactor. Fed with synthetic wastewater (SWW) similar to that used by van De Graaf et al. (1996) at 35°C, the enrichment reactor achieved conversion rates of about 3,000 mg-NH₄-N/l/d. Nevertheless, practical application of the Anammox process is still challenging. One big challenge is the slow growth rate of the microorganisms, with a doubling time of 7 to 14 days causing slow start-up (Strous et al. 1998; van der Star et al. 2007; Ali et al. 2015).

Anaerobic digestion and ammonium oxidation were used mostly for highly polluted wastewaters showing good degradation performances (van de Graaf et al. 1996; Kalogo & Verstraete 2001; Fux et al. 2002; Foresti et al. 2006; van der Star et al. 2007). Adaptation of the microorganisms to low concentration wastewaters is an important challenge in making anaerobic processes suitable for domestic wastewater treatment (Ali et al. 2015; Vega De Lille 2015).

This paper is about the set-up and start-up of a decentralized anaerobic pilot plant combining anaerobic digestion and Anammox-based processes to treat low-strength domestic wastewater. An aim of this project was to produce household service water from domestic wastewater. In contrast to potable water, service water is not clearly defined or regulated. In this project the service water generated meets the requirements of the German wastewater regulations (AbwV 2016) and the Bavarian bathing waters regulations (BayBadeGewV 2008). The treated water should also be aesthetically attractive. The maximum permitted concentrations for ammoniacal-nitrogen (NH₄-N), total nitrogen and phosphorus are 10, 13 and 1 mg/l, respectively. The organic load represented as the COD should not exceed 75 mg-COD/l. As fecal indicator microbes for polluted wastewater, E.coli and enterococci should not exceed concentrations of 900 cfu (colony-forming units)/100 ml and 330 cfu/100 ml (BayBadeGewV 2008; AbwV 2016).

For cost-effectiveness, all three 1,200-liter reactors are cylindrical tanks made of polyethylene. R1 and R3 were designed as batch stirred tank reactors (BSTR), each 1,300 mm tall and of 1,100 mm diameter, giving a filling height-diameter ratio of one. To suspend the microorganisms and homogenize the reactor medium, an axial flow field is induced by stirring in the BSTR. As a result, the suspended bacteria in R1 and R3 must settle before the effluent is pumped to the next process step, in order to guarantee biomass retention. The BSTR outlets are, therefore, about 200 mm above the reactor bottom, providing a retention volume of approximately 200 liters.

Due to the symbiotic nature of the acetogenic and methanogenic microorganisms, a high bacterial density must be ensured in the reactor. R2 is thus a fixed bed reactor (FBR), 2,100 mm tall and 900 mm in diameter. As support media it contains fully 3-D, permeable, BIO-NET^® blocks from Norddeutsche Seekabelwerke GmbH. In the FBR, the bacteria remain attached to the fixed bed and down-stream recirculation through it is essential for good distribution.

The microorganisms used in the process are mesophilic, so the temperature must be maintained at between 30 and 40°C (Grady et al. 2011; Vega De Lille 2015). Thus the wastewater must be heated for good degradation performance. A 1,000-liter buffer tank equipped with a heating cartridge is used for pre-heating. It also facilitates start-up of the three reactors, and offers the possibility of buffering between biological processing and post treatment. It is connected to the collector pipe after the communal sewage plant pre-treatment basins, which remove large particles from the wastewater.

On start-up, the reactors were operated separately, initially, to facilitate microorganism adaptation to their new environments. Then, R1 and R2 were linked by replacing the municipal wastewater (MWW) supplemented with additional organic acids at the outlet from R1. Finally R3 was linked in by substituting the artificial wastewater with the effluent from R2.

R1 and R2 were each inoculated with 60 liters of seeding sludge from a two-stage anaerobic digester in Obermichelsbach (Germany), where vegetable waste products like sugar beet residues are degraded and used for sustainable energy production. Both reactors were fed from the start with mechanically pretreated MWW. Degrading predominantly short chain organic acids by acetogenesis and methanogenesis, R2 needed additional substrate until it was connected to the hydrolytic and acidogenic process (phase I). As suggested by Kuba et al. (1990) and Vega De Lille (2015), acetic, propionic and butyric acids were added to R2 in the ratio 2:1:1. As MWW was used from the start, trace elements were not needed. After adaptation, the connection of R1 and R2 was carried out stepwise (phase II). R2's initial influent was 1,000 liters of fresh MWW supplemented with organic acids–as above–it was replaced by 30, 50 and finally 100% effluent from R1. When reactor connection was complete, no additional substrate was required (phase III).

The pH also affects degradation by microorganisms. Lab-scale studies showed that adjusting the pH to the microorganisms' optimum value affects the intermediate products to improve degradation performance. In hydrolysis and acidogenesis the optimum pH is between 4.5 and 6, whereas for acetogenesis and methanogenesis the optimum is around 6.8 to 7.8 (Grepmeier 2002; Vega De Lille 2015). The pH in the pilot-plant was set at the start of each batch in phases I and II to 5.5 in R1 and 7.5 in R2. In R1 it was adjusted by adding approximately 200 ml of hydrochloric acid (15% v/v), in R2 about 700 ml of caustic soda (15 v/v) were added. Daily operation of pilot-scale treatment involves high consumption of auxiliary materials. For cost-effectiveness, degradation performance was determined without pH adjustment from phase III of the start-up onward.

The Anammox-based process in R3 was inoculated with 60 liters of seeding sludge from a de-ammonification (DEMON) reactor in Fulda Gläserzell (Germany). In phase I, SWW, comprising 1,000 liters of tap water mixed with ammonium sulfate (40 mg-NH₄-N/l) and sodium nitrite (50 mg-NO₂-N/l) was introduced. This was followed by stepwise adaptation to organically degraded MWW from R2 at MWW:SWW ratios of 20:80 (phase II), 50:50 (phase III), and 80:20 (phase IV). In phase V the feed to R3 was 100% MWW. Anammox bacteria are mesophilic and the optimum temperature for them is between 30 and 40°C (van de Graaf et al. 1996; Strous et al. 1998). Further studies, however, demonstrated good degradation at temperatures around 18 to 20°C (Isaka et al. 2007). For this study the temperature in R3 was held between 28 and 35 °C. The pH range for Anammox bacteria is between 6.7 and 8.3 (Strous et al. 1998; Wesoly 2009). As the process step in R2 is in the same pH, no pH adjustment or control is needed.

Once the three reactors had been connected, the pilot plant was tested for 200 days using mechanically pretreated MWW.

Both online and offline measurements were carried out to analyze degradation performance. The fluid temperature in all reactors and the buffer tank was monitored online with Pt100 SITRANS TH400 temperature sensors from Siemens AG, Munich, Germany. As noted, fluid temperature is adjusted only once, in the buffer tank, to 35°C. Thereafter it decreases with time in the process units at an average of 1.125°C/hr.

The pH was monitored in all reactors with pHD-S sc Digital Differential pH-sensors from Hach Lange GmbH, Düsseldorf, Germany. Degradation by the Anammox-based process (R3) was analyzed with a 3798-S sc Digital inductive conductivity sensor from Hach Lange GmbH (Düsseldorf, Germany).

To characterize the organic load in the water, the COD was measured offline using a QuickTOC^® analyzer from LAR AG, Berlin, Germany.

Nitrogen removal was characterized using various parameters, including the concentrations of ammoniacal-nitrogen (NH₄-N), nitrite-nitrogen (NO₂-N) and nitrate-nitrogen (NO₃-N). The determinations were done with photometric-based analytical test kits from Merck Chemicals GmbH (article-no. NH₄-N: 1.00683.0001, NO₂-N: 1.14776.0001 and NO₃-N: 1.14773.0001).

In phase II, the SWW was replaced stepwise with effluent from R1 at 30 (batch 4), 50 (batch 5), and 100% (batch 6). With an influent COD concentration of 343 mg-COD/l and COD removal rate of 141 mg-COD/l/d, removal efficiency of 71% was achieved in batch 4. In batch 5 the COD removal efficiency was 47%, with a starting concentration of 566 mg-COD/l and removal rate of 258 mg-COD/l/d. When the connection of R1 and R2 was accomplished (batch 6), the degradation removed around 410 mg-COD/l/d, with an influent COD concentration of 562 mg-COD/l, resulting in 71% removal efficiency. Thus, while the micro-organic environmental conditions changed, the average removal efficiency in phase II was about 63%.

When R1 and R2 were fully connected, organic acid supplementation ceased (phase III). Although the influent COD concentrations – average 305 mg-COD/l – were lower than in the two previous phases, the COD removal efficiency remained the same at around 62%. With a total COD removal rate of 198 mg-COD/l/d, an average effluent concentration of 112 mg-COD/l was achieved.

Finally, it was shown that start-up of the two-stage anaerobic digestion process was successful at a temperature range between 33.7 and 37.5°C, without the need for specific pH control. The pilot plant achieved COD removal efficiencies of 62% on average. Comparable results (69 and 50%) were achieved by Donoso-Bravo et al. (2009) in a similar lab scale plant, even though higher COD influent concentrations of about 500 mg-COD/l were involved. While the pilot plant's average COD removal rate is lower than the results presented in literature (Donoso-Bravo et al. 2009), the target service water quality of 75 mg-COD/l was almost achieved, even during start-up.

It is thought that the fall in ammonium removal efficiency in phases III to V arose from inappropriate nitrite-nitrogen (NO₂-N) and ammoniacal-nitrogen (NH₄-N) ratios. The ratio of NO₂-N/NH₄-N, as recommended in the literature, should be between 1.15 and 1.3 (van de Graaf et al. 1996; van der Star et al. 2007; Ali & Okabe 2015) After raising the nitrite concentration to give an NO₂-N/NH₄-N ratio in the range 1.15 to 1.3, removal efficiency increased to 84.1, 95.0 and then 97.2% (phase VI), with an average ammonium degradation rate of 189 mg-NH₄-N/l/d.

The Anammox-based process was started successfully within four weeks for treating domestic wastewater at temperatures between 28 and 35°C, and with pH in the range 7.1 to 8.2. Average ammoniacal-nitrogen removal efficiency of 92% was achieved for ammonium-poor wastewater by the end of the start-up. Comparable values of 85 to 99% are reported in the literature for ammonium-rich wastewater (Fux et al. 2002).While the pilot plant's average ammoniacal-nitrogen removal rate was relatively low compared to the results achieved by Fux et al. (2002), the self-defined service water quality limit of 10 mg-NH₄-N/l had already been reached during start-up.

With an average initial COD concentration of about 212 mg-COD/l in MWW, two-stage anaerobic digestion reduced the COD concentration to 85 mg-COD/l, an average COD removal efficiency of 60%. R3 removed another 21% of COD, giving an average effluent COD of 39 mg-COD/l. Total COD removal efficiency of the three-stage plant during 200 days of operation was about 81%. The first two stages could not meet the self-defined service water limit of 75 mg-COD/l on their own. With the Anammox-based reactor downstream, however, the limit was always achieved.

NH₄-N concentrations in the influent MWW were 46 mg-NH₄-N/l on average. During anaerobic digestion the ammoniacal-nitrogen concentration rose on average to 50 mg-NH₄-N/l, which can be explained by the hydrolysis of protein-rich wastewaters (Grepmeier 2002). In the Anammox-based stage, however, it decreased, on average, to 1 mg-NH₄-N/l. This represents an ammonium removal efficiency of about 96% with six hours' retention time, and the self-defined limit of 10 mg-NH₄-N/l was undershot substantially.

This paper is a description of the successful start-up of a decentralized anaerobic pilot plant for treating domestic wastewater to service water standards. At first, all reactors were operated separately to facilitate microorganism adaptation to the new environments.

The initial, two-stage anaerobic digestion was operated in a temperature range of 33.7 to 37.5°C, without pH-control. Even in this phase, an average COD removal efficiency of 62% was achieved, reducing the average influent COD concentration of 305 mg-COD/l to 112 mg-COD/l. Average COD removal efficiencies of 60% in the subsequent, 200-day, operational testing phase show that the process was stabilized successfully. While achieving an effluent COD concentration of approximately 85 mg-COD/l, the two-stage treatment could not satisfy the self-defined service water quality limit of 75 mg-COD/l on its own. With treatment by the Anammox-based reactor downstream, however, the limit was undershot (39 mg-COD/l).

Ammoniacal-nitrogen removal efficiencies in the Anammox process were between 90 and 79.8%, during feeding with 100% SWW. Substrate limitation led to low removal efficiencies during adjustment to MWW. The optimum NO₂-N/NH₄-N ratio for the Anammox-stage of the pilot plant was determined as 1.14 and the nitrite feed was adapted to this. Subsequently, ammonium removal in 100% MWW was about 95% and the self-defined limit for service water of 10 mg-NH₄-N/l was achieved with a retention time of six hours. With an average removal efficiency of about 96% during the subsequent 200 days of operation, it is safe to say that the Anammox-process was stabilized successfully.

The authors acknowledge gratefully the Hans-Sauer Foundation for financial support for setting-up and starting-up the pilot plant; the communal sewage plant in Erlangen, Germany, for their help, and SIEMENS AG, Germany, Schraml GmbH, WEBFactory GmbH, ZWT Wasser- und Abwassertechnik GmbH, Maschinenbau Biermann GmbH, Albert Handtmann Armaturenfabrik GmbH & Co, and Norddeutsche Seekabelwerke GmbH for the support and equipment provided. H. Kang is also thanked for her reliable contribution to the work during the start-up phase.