A full-scale hybrid vertical anaerobic and aerobic biofilm wastewater treatment system: case study

High organic content wastewater is inherent to many industrial activities, especially the food and beverage industries. With increases in farm productivity and demand of local and international food markets, food processing industries are developing fast. Challenges that accompany this development are increased waste water/organic production that need to be treated in both economically and environmental friendly manners. The challenges are especially noted in Norway, where local industries' wastewater is normally discharged directly to municipal wastewater treatment plants (WWTPs) for treatment. Considering the fast growth of industries versus the relatively stagnant WWTP upgrading pace, increased water discharges normally overload the WWTPs, resulting in high fees and even penalties for the local industries. Another challenge typically in Norway is that WWTPs constructed in the late 20th century were dominated by physico-chemical processes where organics are mainly removed by particle separation and the residuals and nutrients are handled by coagulation and sludge handling afterwards. Thus, those WWTPs struggle to meet stringent discharge requirements with solely non-biological processes. An urgent need for industries, WWTPs and communities is to find solutions to solve these immediate challenges.

An industry on-site biological wastewater treatment process with small footprint and high efficiency can fit the niche for overcoming the above challenges. Using biological processes for wastewater stabilization has been well known and established. Anaerobic digestion (AD) as a broadly recognized sustainable wastewater treatment method has been the core of high organic content wastewater treatment. Highly efficient AD (Lim & Kim 2014), such as expanded granular sludge bed (EGSB) and internal circulation (IC) requires less space, have a low sludge yield and produce renewable energy as methane has gained fast development (Tchobanoglous et al. 2013). However, to meet stringent effluent requirements, anaerobic digestion needs to cooperate with post-polishing steps for example an aerobic process (Tchobanoglous et al. 2013). A compact aerobic stage such as a moving bed biofilm reactor (MBBR) (Bassin & Dezotti 2018) or continuous flow intermittent cleaning (CFIC^®) reactor (Rusten et al. 2011; Wang et al. 2017) developed for organic/nutrients removal can be a good post-treatment option. A successful integration of the two high rate systems, for example EGSB with MBBR/CFIC, into one compact vertical system such as a HyVAB^® (Wang et al. 2017), can reasonably give a series of potential benefits to the industries such as high organic removal efficiency, small footprint and energy recovery.

The first full scale HyVAB^® that comprises a bottom EGSB stage and an aerobic biofilm stage (CFIC^®) stacked in one compact reactor was installed at a vegetable processing facility in late 2016 in Grimstad, Norway, and started operation in February 2017. The full-scale plant is dedicated to the wastewater generated in the factory where different vegetables are processed throughout the year. In this study, the performance of the HyVAB system in different production seasons was recorded and presented. The COD removal efficiency, biogas generation and sludge yield were assessed. The operational cost and energy balance was analyzed. This study will provide design and operational reference for upgrading and deployment of HyVAB technology in similar industries.

The case treatment system consists of a pretreatment tank for sand removal, an equalization tank for wastewater storage and pH adjustment, a HyVAB reactor, a sludge settler and a flare system for biogas combustion (Figure 1). The full-scale plant is fully automated and can be remotely controlled through a software interface, Teamviewer.

The full scale HyVAB reactor integrates a EGSB reactor (96 m³) with the CFIC stage (30 m³) (Figure 1). The reactor has a cross sectional area of 10 m² and a total height of 14 m. The two stages were separated by a three-phase separator. Generated biogas in the AD stage was collected by the three-phase separator and further ventilated throughout the system. In the aerobic stage, BWTX^® and BWT15^® (Biowater Technology AS) carriers (surface to volume ratio of 650 and 828 m²/m³, respectively) were used as biofilm attaching substratum. Air was supplied from a blower to the bottom of the aerobic stage for organic consumption and biofilm growth.

The food processing factory produces different seasonal vegetables and the production season runs throughout the year, with five-weeks' break during summer. Vegetables that were being processed included potatoes, cucumbers, carrots, green peas, red cabbage, beets, etc. The vegetable processing wastewater's quantity and quality in terms of organic content, pH, solid content, etc. varied corresponding to different production seasons. A recorded yearly average of the wastewater total nitrogen (TN), total phosphorus (TP) and conductivity were 23 and 4.7 mg/L and 61 ms/m, respectively. The wastewater from the production line went through a 3 mm rotation sieve before entering the sand trap (Figure 1). Wastewater was pumped from the sand trap intermittently to the equalisation (EQ) tank for pH correction, giving sufficient organic acidification before feeding to the HyVAB. Caustic soda was dosed to the EQ tank until it reached pre-set pH values.

Analysis showed that the EQ tank wastewater total COD was in the range of 300 to 6,700 mg/L, depending on the processed vegetables. The wastewater pH before entering the EQ tank was from 3.8 to 6.0 and was maintained at pH 5–5.2 in the EQ tank due to wastewater acidification and pH adjustment. The total and volatile suspended solids (TSS and VSS) contents were from 30 to 2,100 mg/L and 30 to 900 mg/L, respectively.

The plant was started for operation in late February 2017. The recorded data is up to early November 2017. A total of 89 operational data points were recorded. Before February 2017, production wastewater was discharged directly to municipal sewer system. The factor processes similar products from one year to another. During the summer break, when the facility suspended operation, the reactor was shut down and a minimal aeration continuously applied in the aerobic stage to maintain basic bacterial activity.

The HyVAB plant was running at a relatively stable temperature of 25 °C during this period. Steam injection to the sand trap was applied to maintain a stable temperature in the wastewater during cold seasons.

Granular sludge, with a particle size of ∼2 mm, from Netherland was inoculated in the AD stage of HyVAB. Approximately 50 m³ (70 kg VSS/m³) of the granular sludge was seeded to the reactor at the commencement of the test.

18 m³ of plastic biocarriers BWTX^® and BWT15^® (Biowater Technology AS) were added in the aerobic stage, giving an approximate surface area of 12,300 m². Air was supplied at a rate of 400 ± 100 L/h to maintain a dissolved oxygen (DO) level of 4 ± 1 mg/L in the aerobic biofilm stage. The aerobic CFIC stage was running in washing mode as a moving bed biofilm reactor (MBBR) throughout the operational period (Bassin & Dezotti 2018).

The wastewater was acidified to an average 40% of total feed COD, by controlling pH level in the EQ tank. A start up loading rate was 10 m³/h and went up and down corresponding to different operational seasons. The maximum loading rate has been tested was 14 m³/h, giving an AD stage hydraulic retention time (HRT) of seven hours and a total HRT of 10.3 h. Recycle was applied from above the three-phase separator back to the bottom of the AD stage at rates from 16 m³/h to a maximum 25 m³/h, giving up-flow velocities from 1.6 to 2.5 m/h. The maximum organic loading rates to AD stage reached 17 kg COD/m³ d.

Nutrient (NP5^®, NH₄NO₃ with N: P = 5:1, from Yara) were added to the HyVAB recycle line to provide necessary nutrients for organisms' synthesis. Caustic soda was also dosed continuously in the recycle line to neutralize feed water to optimal pH.

Regular liquid samples were collected from the EQ tank, above the three-phase separator and effluent of the HyVAB system for monitoring the reactor's performance. The granular sludge and biofilm developments were also monitored intermittently. Total and soluble COD, pH, volatile fatty acids (VFAs), TSS and VSS were measured for each obtained sample while ammonium, orthophosphate, biogas contents and alkalinity were measured occasionally.

TS, VS, TSS and VSS (filtered with a 1.5 μm pore size glass filter) were determined based on the standard methods (APHA 2012). The alkalinity was measured by potentiometric titration to end-point pH (APHA 2012). Commercial kits (Hach Lange) were used to determine VFA and total and soluble COD content (filtered right after sampling with a 0.45 μm pore size glass filter). pH and temperature were measured using a VWR pH110 and DO was measured using a WTW Oxi 3315. Biogas composition was analyzed by a BIOGAS 5000 (Geotech).

The full scale HyVAB was started up after seeding granules to the AD stage in late February 2017. The results are presented below.

Figure 2 shows the feed and effluent total COD and the removal efficiency during the operational period. Feed total COD was generally around 2,600 mg/L before May 2017 when potatoes were processed in the factory. The feed COD varied from over 3,000 mg/L to less than 500 mg/L after May when fed with green beans, cucumbers, etc. wastewater. It shows that, overall, 90% of total COD in the feed was removed during this period, with some periods when efficiencies went up and down corresponding to the load fluctuations (Figure 2). The effluent TCOD was on average 268 mg/L during the whole operational period, which was about half of the discharge requirement of 500 mg/L.

Feed wastewater TSS was generally under 1,000 mg/L, with some exceptions over 1,000 mg/L (Figure 3) when potatoes were processed with dirt mixed in the wastewater through the sand trap. Generally, potato wastewater contains TSS over 600 mg/L (before May 2017, Figure 3). Green beans were processed before July, giving a TSS around 400 mg/L. Cucumbers were processed during August, giving a relatively low feed TSS of around 200 mg/L (Figure 3). Red beets were processed after 10^th October to the end of October, where a high TSS was observed. During this period, some other vegetables were also processed but at a small quantity and in short periods, which are not specified here.

Effluent TSS values were obtained from measurements of samples to the municipal WWTP (Figure 3). TSS removal efficiency in Figure 4 shows that on average over 60% of the feed TSS was reduced, giving an overall value less than 200 mg/L effluent TSS (Figure 3). The sludge settler was also applied during this period to test the sludge removal efficiency. The effluent flow rates through the settler was in the range of 0.4 to 1.3 m/h with a HRT from 1.5 to 0.5 h. It shows that the TSS removal efficiency was on average 50% when the settler was not in use (from 10^th March to 11^th April and June, Figure 4). 70% of sludge was removed when using the sludge settler for HyVAB effluent (from 11^th March to 10^th April and after 10^th August, Figure 4). The sludge settler contributed about 20% TSS removal in the effluent wastewater. Higher effluent TSS was observed when the sand trap and the sludge settler were not emptied in due time at the end of the operational period (Figures 3 and 4). The observed overall biomass yield was 0.11 ± 0.01 g VSS/g COD removed.

The effluent NH₄-N and PO₄-P were recorded on average lower than 25 and 10 mg/L, respectively. To minimize the nutrient consumption, nutrient dosing was controlled based on flow and organic loading rate to the reactor. The NH₄-N and PO₄-P were added to the reactor at a COD ratio of COD: NH₄-N: PO₄-P = 500:5:1. No discharge limit was set for these two parameters in this treatment plant.

Biogas production was recorded intermittently with an on-site biogas analyzer, which showed that methane accounted for an average 84% of the generated biogas volume. This value is in accordance with the pilot scale HyVAB test results (Wang et al. 2017). Other detected gases were CO₂, nitrogen and H₂S. H₂S concentration was shown to be less than 200 ppm. Burning of biogas containing H₂S will generate SO₂. To use the biogas for steam generation, a H₂S concentration lower than 200 ppm will be required, otherwise H₂S removal will be required.

A real-time biogas flow rate recorded through a biogas flowmeter is shown in Figure 5 for the period from late March to early April 2017. A mass balance (Table 1) shows that during this period an average 82% of feed COD was converted to methane in biogas. The total COD removal efficiency through HyVAB was over 90% (the sludge settler was not in use during this period). It indicates that the aerobic stage removed about 10 to 12% of the residual COD that was not consumed in the AD stage.

The methane yield was on average 0.39 L/g COD removed during this period. The overall methane yield is given in Figure 6. Some high levels of yields were observed at the start of operation after each production break. The release of biogas from the feed and recycling agitating effects in the anaerobic stage led to the high methane yield values.

The anaerobic stage granular sludge was monitored in April and June for its size, activity and content. It shows that after four months' operation, the granules were getting smaller from an average 1.44 mm to 1 mm and the settling velocity reduced from 82 to 63 m/h. The sludge settling rate was still within a good range for a high rate anaerobic stage. Initial seeded sludge was over 80% of the AD stage volume to account for sludge lost due to start up and new environment acclimation. The TS content was recorded to be 5.2 tons at the commencement of the operation. Sludge lost was observed during the operation, while analysis showed that the active part (organic dry matter) increased by 5% in the granular sludge compared to the original seeded sludge, indicating increased sludge activity (Xu et al. 2015). The granular sludge loading rates in HyVAB were lower than 0.4 kg COD/kg VS·d during the whole recorded period.

The original granules were from paper and pulp wastewater treatment. This wastewater possesses comparably higher water hardness than the vegetable processing wastewater. The size and settling velocity reduction of the granules can be attributed to the low feed wastewater hardness and its relatively easily degradable nature.

The aerobic biofilm stage worked as a polishing step for liquid flow from the anaerobic stage. Feed COD was consumed mostly (80%) in the AD stage, leaving less than 20% of the feed total COD to the aerobic biofilm stage (Figure 2 and 3). The surface loading rate to biofilm reached a maximum 20 g COD/m² d. A DO concentration was kept above 4 mg/L during the operational period to provide sufficient oxygen for biofilm growth. Biofilm on the carriers were recorded intermittently, which showed an average of 12 g TS/m². The biomass content on carriers were lower than previously tested at lab scale, which was on average 20 g TS/m². A strong agitation from aeration with an upflow velocity of 45 m/h constantly scoured and washed out detached biofilm from the reactor.

The production seasons give different wastewater content and wastewater quantity, while the HyVAB design allows for a certain flexibility to handle high load situations. The HyVAB experienced relatively stable hydraulic loads with maximum organic loads within the design limitations. No overloading situations were experienced. The decrease of load organic will give low methane production without affecting the system performance. Inoculated granules were able to handle loads ranging from high to low organic load and even periodical plant shutdown due to lack of wastewater.

The HyVAB system has been proven to be highly efficient in reducing medium to high levels of COD from food industry and petrochemical wastewater (Wang et al. 2017). The integration of a high rate AD stage with a compact moving bed biofilm system provided good COD removal efficiency. The AD stage performance was enhanced by this integration feature, showing higher biogas methane content (over 84%) and consistently high efficiency (Wang et al. 2017). The system was also able to restart quickly after a long summer break and intermittent stops due to lack of wastewater. COD removal was consistently high with good biogas generation observed within six hours of process restart. High quality granules were important for such a resilient system and stable HyVAB performance. Granule sludge was checked after one year of operation. Granules were able to be retained in the system with intermittent on/off operation and the heavy sludge morphology has not been drastically changed. The moving bed biofilm stage retained active biofilm and generate sludge that had good settling ability (SVI less than 100 ml/g TSS). The long term and intermittent food scarce may have enhanced the biofilm attachment, when the system restarted, the biofilm activity recovered fast. The complete system biomass yield combining sludge settler was less than 0.1 kg VSS/kg COD removed. Further solids removal from such effluent water by for example using membrane filtration can promote water reuse on industry sites.

Due to the close integration of the anaerobic and aerobic process in a compact system, the dissolved gases (methane, H₂S etc.) generated in the AD stage were not released to atmosphere, avoiding a commonly observed emission problem in anaerobic treatment plants (Mannina et al. 2016). CO₂ was stripped out from the water in the biofilm stage.

The holistic plant consists of operational room, chemical room, equalization tank, HyVAB, gas flare and baffle system and sludge settler and occupied a space of 546 m² (15 × 36 m). The energy consumption mainly included aeration, pump and lightning. Electricity consumption was approximately 0.5 kwh/m³ wastewater treated. The generated biogas energy content was close to 3,838 kWh/d, assuming at design load of 1,440 kg COD/d and a COD conversion ratio of 0.82 (AD stage) and methane yield 0.39 L/kg COD removed. This energy is potential substitute to the factory heat demand.

Successful startup and operation of the full-scale hybrid vertical anaerobic biofilm reactor (HyVAB) treating vegetable processing wastewater was achieved at varying organic loading rates for eight months. The unique features of HyVAB with close integration of anaerobic and aerobic stages enable over 90% total COD removal with low sludge production (0.1 kg VSS/kg COD removed) and 82% of feed COD converted to methane. The produced biogas contained 84 ± 2% methane and discharges to air and the treated water were odorless. The plant was proved to be a compact and energy efficient system for wastewater treatment on site.