Prefiltration pilot study with drum filters

The Rya wastewater treatment plant (WWTP) in Gothenburg, Sweden, is facing new and stricter effluent requirements, and at the same time, a continued increase in population is expected in the coming years. Expansion of the current WWTP on new areas is necessary and a project, Nya Rya, started in 2021 which aims to complete this expansion and be fully operational by 2036. The project is also evaluating the capacity of the current WWTP and how to co-exist with the new processes. A technology screening in 2021 led to the decision to run a pilot study during 10 months with a drum filter as a possible efficient pretreatment step in addition to the conventional pre-settling tanks. The pilot study included chemical dosing of both polymer and polyaluminum chloride (PAC) prior to the drum filter, testing periods with higher suspended solids (SS) loads in the influent and two different pore sizes (100 and 300 μm).

The aims of this pilot study were to conclude if micro screening in drum filters is a viable complement or alternative to the existing conventional pre-settling tanks, to gain knowledge of reduction of organic materials and phosphorus (P) at different loads, give input to the Nya Rya project for possible dimensioning of pretreatment steps and to gain practical experience with drum filters in terms of operation and maintenance.

Surface efficient complements or alternatives must be considered if the current pretreatment at Rya WWTP requires increased capacity in the future or in case parts of the area currently occupied by the pre-settling tanks would be needed for other processes. The pretreatment alternatives also need to be able to handle momentaneous peaks in SS and flow during tunnel flush events. Micro screening in drum filters could be a potential pretreatment compliant at the Rya WWTP due to its smaller footprint. Previous studies have concluded that different types of microscreens can be competitive alternatives to conventional pre-settling, especially if available space is an issue (Ljunggren 2006).

The Rya WWTP is a regional plant that treats wastewater from Gothenburg and seven nearby municipalities with approximately 800,000 people connected. The plant was originally commissioned in 1972 as a high loaded activated sludge plant and later expanded with pre-settling, increased aeration and simultaneous precipitation with ferrous sulfate in the 1980s. The plant was expanded for nitrogen removal with pre-denitrification in the activated sludge and nitrification in trickling filters in the late 1990s. Starting in 2005, parts of the pre-settling tanks are used as parallel chemical treatment during high loads when the total influent flow is greater than the capacity of the biological treatment steps. The plant was further expanded with post-denitrification in a moving bed biofilm reactor (MBBR) and microscreening in 15 μm disc filters as a final polishing step in 2010. During 2017, additional MBBR reactors were taken into operation for post-nitrification and for sludge liquor treatment, by deammonification.

The current 12 conventional pre-settling tanks make up a total volume of 22,500 m³ and an area of 5,800 m². The influent water passes through 20 mm bar screens, a sand trap and finally 2 mm step screens before reaching the conventional pre-settling tanks. Six out of 12 pre-settling tanks can be used for direct chemical precipitation with polymer and PAC to avoid overloading the biological treatment steps during rain weather flows. Tunnel flush events are performed weekly to flush sediment in the tunnel system close to the WWTP. These events cause momentary higher peaks of SS and flow to the plant.

The drum filter pilot was a HDF1604 (Hydrotech Veolia) and the total tested surface area was 2.4 m². The installed pilot filter area was too large compared to the rest of the pilot system, so only one-third of the total available area in the pilot was tested in this study. Therefore, it was determined by the supplier prior to the pilot trials that two-thirds of the filter area needed to be plugged to keep the hydraulic retention time (HRT) of the chemical precipitation within the correct time interval to achieve enough coagulation prior to the filter. Using two-thirds of the filter area would correspond to the provided design of the hydraulic capacity of the drum filter that the study aimed to evaluate. Using the whole available filter area would result in operation with too low backwash frequencies resulting in no possibility to test the filter in the design operating conditions. Two different pore sizes for the filter elements were tested, 100 and 300 μm.

The study used a cationic polymer (H6358 from Hydrotech) and PAC (PAX XL 100 7.5% Al from Kemira). A cationic polymer was chosen over an anionic one based on the recommendation from Hydrotech Veolia who had prior experience with drum filters where cationic polymers resulted in stronger flocs and clearer effluent. A dosage of 10–15 mg Al³⁺/L is normally required to reach a total P concentration of 0.2–0.3 mg/L in the effluent according to prior experience (Hydrotech Veolia). Dosing set-points in the range of 5–8 mg Al³⁺/L were used during this pilot trial to see how close to this range effluent P concentrations the drum filter would be able to achieve. The dosing set-point for the polymer was set in the range of 3–5 mg/L.

Two different modes of operation were tested, a level-controlled mode and a flow-proportional mode. The level-controlled mode adjusted the influent flow to keep it as high as possible to maintain a level set-point inside the drum (390 mm). This mode did not have any chemical precipitation and no bypass was possible since the level inside the drum was set. The flow-proportional mode controlled the influent flow to the pilot based on the influent flow to the full-scale Rya WWTP. A scaled signal from the Rya WWTP's process control system adjusted the influent flow to the pilot. The chemical precipitation to the drum filter started automatically at a typical rain weather flow, which was determined to be above 6.5 m³/s to the Rya WWTP, corresponding to influent flows to the pilot above 73 m³/h and surface loading rates approximately 31 m³/(m², h). Direct chemical precipitation of the conventional pre-settling tanks usually starts at influent flows around 8–10 m³/s to the Rya WWTP. Influent flows above 6.5 m³/s was chosen over 8–10 m³/s to get longer periods with chemical precipitation to evaluate in the pilot study. Five different chemical precipitation campaigns were carried out during the 10 months pilot study in total testing different dosing set-points, flocculation with only polymer and chemical precipitation with both polymer and coagulant.

Weekly 24-h composite samples were taken by automatic samplers from the influent and the effluent at rotating days. If the sample was taken on a Tuesday one week, the sample was taken on a Wednesday the following week. The composite samples were not flow-proportional. Weekly grab samples were taken from the sludge effluent. Both total and filtered fractions were analyzed; total fractions were analyzed without any handling prior to the analysis and filtered fractions were filtered through a 0.45 μm filter (Table 1).

SS, PO₄-P, TS and VS were analyzed by the internal laboratory at the Rya WWTP, and the remaining analyses were sent to an external laboratory (Eurofins). The samples to be analyzed externally were stored in cooled containers and sent for analysis the same day. The analytical methods used at both laboratories were according to the Swedish Institute of Standards (Table 2).

Sampling campaigns for chemical dosing of polymer and coagulant as well as tunnel flush events, (periods with short but very high influent SS), were conducted by taking 12 grab samples during a 24-h period. These grab samples were then analyzed the same way as the regular 24-h composite samples. The chemical precipitation campaign started dosing of both polymer and coagulant in the afternoon the day before the first sample was taken to prevent any initial disturbance, the dosing set-point for the polymer set to 3 mg/L and the dosing set-point for the PAC was set to 5 mg Al³⁺/L. Tunnel flush events are done routinely at the Rya WWTP to remove as much as possible of grit and other various suspended particles that sediments in the tunnel system. The sampling started a few hours before the start of the tunnel flush and continuously sampled every other hour for 24 h. Both sampling campaigns used the level-controlled mode of operation.

The pilot had two level sensors (one inside the drum and one outside), two turbidity sensors (one for the influent and one for the effluent) and a combined temperature/pH sensor. Each sensor was cleaned weekly my manual wiping and spraying with mild detergent. The backwash strainer was removed and cleaned weekly or more often based on need. The backwash nozzles were visually inspected weekly and cleaned when needed.

Filter elements will clog up over time and needs to be washed regularly to maintain capacity. The filter elements in the pilot were manually cleaned monthly with a pressure washer that sprayed clean water (drinking water quality) on the outside of the filter elements. The filter elements were also cleaned chemically every 3 months. A service technician (Hydrotech Veolia) performed the chemical cleaning with 4–5% hydrochloric acid (HCl) followed by 1–2% sodium hypochlorite (NaClO). HCl was sprayed first after which the first automatic washing was started. NaClO was sprayed after the first automatic washing was completed and a second automatic washing was started. A risk assessment of the chemical washing was performed (Hydrotech Veolia). Full-scale installations are equipped with automatic chemical washing systems.

The maximum capacity of the drum filter was tested during the level-controlled mode, where the water level inside the drum filter was kept constant. The maximum influent flow, around 75 m³/h, was similar for 100 and 300 μm at lower concentrations of SS in the influent but 300 μm was able to maintain higher flows as the SS concentration increased during test periods with level-controlled mode. The level-controlled mode kept the influent flow as high as possible to maintain a level set-point inside the drum. The influent flow was then at the maximum possible without any bypass or chemical precipitation. Larger pore sizes in the filter elements took longer to clog so the flow capacity was larger with larger pore sizes. The flow capacity is also related to the open area fraction which can differ depending on the thickness of the threads. The 100-μm filter was able to reach the same maximum flow capacity as the 300 μm but would likely require too frequent chemical washing to maintain the capacity and would hence not be sustainable long term. The average SS concentration in the influent was higher overall during the test periods with 300 μm compared to the period with 100 μm (Table 3).

A large filter capacity is needed to avoid bypass completely, which would result in an unreasonable amount of filter units. Therefore, it is important to consider the degree of bypass at specific times and what level would be acceptable to momentarily load subsequent biological treatment steps. Daily average data did not capture bypass events caused by momentaneous peaks in either influent SS concentration or flow. Hourly data would then be important during design when calculating how much of the flow to the drum filters would be reasonable to bypass.

The pilot trials proved that the HDF1604 drum filter (Hydrotech Veolia) was able to reduce organic materials and P at different nutrients and flow loads, the reduction was similar between 100 and 300 μm in pore size for the influent water tested. The effluent SS concentration was similar to the conventional pre-settling tanks at the Rya WWTP and the trials provided practical experience with operation and maintenance of the filter. The filter, 100 and 300 μm in pore size, had similar maximum flow- and surface load capacities at lower influent concentrations, but 300 μm was able to maintain that capacity for longer and at increasing influent SS concentrations. The sludge from the drum filter had lower TS than the primary sludge from the conventional pre-settling tanks. Chemical precipitation with polymer and coagulant was able to significantly increase the reduction of both SS and P. Chemical precipitation prior to micro screening can then be a way to further reduce particles, if necessary, to limit area requirement as the flocculated particles in the wastewater allows the drum filter to maintain higher hydraulic capacities and without clogging up. The drum filter was able to handle higher peaks in influent SS during tunnel flush events while reducing the SS to similar effluent SS concentrations as during dry weather flow. High flow or SS peaks, either in combination or separately, can cause momentary bypass of the filter like as for the conventional pre-settling tanks during dry weather flow. Daily average data did not capture bypass events caused by momentaneous peaks in either influent SS or flow. Hourly data analysis is needed to correctly design and dimension a pre-filtration process with drum filters.

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

The authors declare there is no conflict.