Large scale tertiary filtration – results and experiences from the discfilter plant at the Rya WWTP in Sweden

In 2010 the Rya wastewater treatment plant in Gothenburg was extended with a moving bed biofilm reactor (MBBR) for post-denitrification and a discfilter plant for tertiary filtration of the effluent in order to be able to meet the new discharge limits of 10 mg/l total nitrogen and 0.3 mg/l of total phosphorus. The disc filters receive effluent from both the MBBR and the secondary settlers. The design of the disc filter plant was preceded by extensive pilot tests where disc filters with a cloth pore diameter of 15 μm was found most suitable for suspended solids (SS) removal (Mattsson et al. 2009).

Microscreens have been used for tertiary filtration since the 1950's when the first drum filters were installed in the UK (Isaac & Hibberd 1972). Since the late 1990's disc filters, rather than drum filters, have been increasingly applied for tertiary treatment, mainly due to small foot-print and low operational head loss. Recent studies have also demonstrated the possibility to combine the technology with preceding coagulation/flocculation in order to achieve <0.1 mg/l of total phosphorus (Tooker et al. 2010; Langer et al. 2012; Väänänen 2014).

The Rya WWTP serves about 830,000 pe with an average flow of about 350,000 m³/d which makes it one of the largest treatment plants in Scandinavia. After screening, grit removal and primary clarification water is treated in a high-loaded activated sludge plant with possibilities for pre-denitrification and simultaneous precipitation. Trickling filters are utilized for subsequent nitrification and the newly installed MBBR for nitrogen removal in a post-denitrification process. Water is finally treated in the disc filter plant before being released into the recipient. The 32 disc filters with a total filtration area of 3,584 m² (Hydrotech HSF 2220-2FN) and with 15 μm pore openings are designed for a flow of 8 m³/s. The filter media is made of a twill weave monofilament polyester filter cloth (Persson et al. 2006). Backwashing requires in the order of 3% of filtered water at design conditions. The streams from the secondary settlers and from the post-denitrifying MBBR are mixed before entering the disc filter plant. However half of the filter units (those to the north) receive mainly effluent from the MBBR whereas the other half (to the south) receive mainly effluent from the secondary settlers. This enables studies of the efficiency of removing particles from the two process steps. Depending on the flow and mass load of SS from the secondary settler and post-denitrification MBBR the operational conditions of the disc filter plant vary; the mass load of SS determines the number of filters in operation and the flow to each filter depends on the resistance over the filter cloth (Table 1). At regular intervals the filters are cleaned with hydrochloric acid (once a month) and hypochlorite (once every three months). The degree of fouling of the filter cloth is measured as the time it takes for tap water to pass the filter cloth compared to passage through cleaned or new filter cloths. The separated sludge flocs are brought back to the inlet channel of the activated sludge tanks.

The objective of this paper is to summarize full-scale results from the first years of operation. In addition, detailed investigations of removal of organic material and nutrients as well as pathogenic microorganisms over specific disc filters receiving mainly effluent from either the secondary settlers or the post-denitrification effluent were performed.

In 2012 removal efficiencies of a series of indicator organisms, including the faecal indicator bacteria E. coli and total coliforms (Colilert-18 Quantitray/2000, IDEXX), and enterococci (membrane filtration ISO 7899-2:2000), were analysed. Besides Somatic coliphages (ISO 10705-2), Clostridium Perfringens (spore forming bacteria, ISOP/CD 6471-2) and the pathogenic virus Norovirus (qPCR) were analysed. Sampling was performed during normal operational conditions. Before the first sampling occasion (6–8 March 2012) each filter was cleaned with acid the day before so that the biofilm build-up was limited. The second sampling occasion (28–30 May 2012) was performed two weeks after the last acid cleaning to see if this would lead to higher removal efficiency due to biofilm build-up. During both sampling occasions the removal of SS over the disc filters were monitored, and during the first sampling occasion also total phosphorus, total nitrogen as well as particle number and size were analysed.

The removal of indicator microorganisms was different depending on type of microorganism. In most samples, the numbers of indicator microorganisms were larger in the effluent from the secondary settler compared to the MBBR effluent (Table 3). There was a slight reduction in numbers of total coliforms and E. coli in the 24-h samples in both measurements for the disc filter receiving mainly MBBR effluent (Table 3). However, less good removal was observed for the grab sample measurement. For the disc filter receiving mainly secondary settler effluent the removal was less good or even negative, with the exception of the enterococci measured with grab samples with an 82% removal efficiency. For Noroviruses and Somatic coliphages, negative removal efficiencies were observed. Clostridium Perfringens were removed in the discfilter receiving both mainly MBBR and secondary settler effluent with removal efficiencies of 78 and 17%, respectively. The removal efficiency over the entire discfilter plant was calculated from the numbers in the part of the disc filters receiving mainly MBBR and secondary settler effluents, respectively, multiplied with the corresponding flows and compared with the effluent concentration from all filters multiplied with the total flow. This showed a substantially higher number of indicator organisms in the secondary settler effluent compared to in the MBBR effluent but that that concentrations in the total effluent was much lower with removal efficiencies >60% (Table 3). One possible cause for the large variation in results for the different sampling occasions can be problems with sample homogenisation considering the large variation in particle size distribution of the different waters. Some larger particles break up during the passage of the filter cloth leaving a larger number of small particles which are easier analysed for indicator and pathogenic microorganisms, especially when colony counting methods are applied. Considering the large variation in data between sampling occasions, disc filters cannot be considered suitable technique for removal of pathogens. Wilén et al. (2012) showed that the reduction of indicator organisms is substantially reduced in the activated sludge process, which indicates that the potential for further removal in disc filters is limited.

Since the discfilter plant was placed in operation 2010, the operational strategies have been improved. From 2012 the filter clothes have been cleaned regularly as opposed to previously when they were cleaned when the hydraulic resistance reached a certain level. The rotational speed of the filters has been increased and hence the frequency of backwashing (Table 1). This has increased the hydraulic capacity of the discfilter plant and allows a higher load of SS which also leads to increased capacity of the secondary settlers.

Frequent cleaning of the filter cloth is important not only for maintaining the capacity (and thereby avoiding bypass) but also to minimize energy consumption. The energy consumption is mainly related to pressurization of filtered water to approximately 7 bar. With reduced capacity, as a result of long-term fouling of the filter cloths, backwashing frequency increases, which in practice means that more filters are needed to treat the same flow. Thus, by keeping the filter cloths clean the production of pressurized filtered water for backwash can be minimized and thereby also the energy consumption. Energy consumption is 0.013 kWh/m³ (Nunes et al. 2013), which is a minor contribution to the overall energy consumption at the plant. From the same study it can be concluded that the cost for chemical cleaning, on a yearly basis, is very small in comparison to the cost for electricity for the backwash pumps. Therefore regular and frequent backwashing is of outmost importance to keep backwashing at minimum and to maintain the design capacity.

In conclusion, the discfilter plant has performed well since start-up maintaining phosphorus concentrations below 0.3 mg P_tot/l and SS below 5 mg/l. Also the secondary settlers can be loaded a bit higher due to the efficient particle separation in the discfilters which increases the efficiency of the whole plant. This study demonstrates successful operation of a large discfilter plant with large variations in flow and particle loading.

The authors thank Christine Enerhall and Emma Stenmark for assistance with experimental work and the staff at Gryaab AB for help with sampling and experimental equipment.