MIXED LIQUOR SCREENING TO PREVENT FOULING IN A MEMBRANE BIOREACTOR WASTEWATER TREATMENT PLANT

Transcription

1 MIXED LIQUOR SCREENING TO PREVENT FOULING IN A MEMBRANE BIOREACTOR WASTEWATER TREATMENT PLANT Nicole Hughes 1, Luke Moran 1, Alice Connell 1, Phil de Groot 1 1. TRILITY Pty Ltd, Adelaide, SA, Australia ABSTRACT The Victor Harbor membrane bioreactor plant has suffered significant operational issues caused by the agglomeration and braiding of fibrous materials. Previously, rectification was achieved by removing and physically cleaning the membranes, which was a costly and high risk process. This paper discusses the different options trialled and implemented by TRILITY to reduce the concentration of fibrous material. Mixed liquor screening was found to be the most effective method with pilot trials successfully removing solids at a rate of 17kg/m2/hr and improving filterability of the screened mixed liquor. A full scale solution has since been implemented at the site and is currently undergoing optimisation with expected long term results of increased membrane life and reduced frequency of physical cleans. INTRODUCTION TRILITY operates the Victor Harbor Wastewater Treatment Plant (VHWWTP) in South Australia under a Design, Build, Operate and Maintain (DBOM) agreement with SA Water. The submerged membrane bioreactor (SMBR) treatment process has an ultimate capacity of 5.1 MLD and utilises Kubota flat sheet membranes in a double deck formation. Plant operations commenced in 2005 and since this time the agglomeration and braiding of fibrous material in the mixed liquor has caused significant operational challenges. This site has experienced issues in achieving the required membrane flux, reduced membrane life and high treated water turbidity at times due to premature membrane failure. In August 2008, plant operations reached a critical point which required all membranes to be removed from tanks and physically cleaned. It was during this process that the full extent of the ragging and fouling across the membranes was realised with the majority of the top deck membranes rendered inoperative. Figure 1 illustrates the degree of braiding that was experienced in the plant with a comparison between heavily braided and clean membrane permeate tubes. The cotton braiding experienced in the plant originates from smaller fibrous suspended solids, commonly found in sanitary products, and cellulose fibres which enter the plant. Under the highly agitated conditions of the bioreactors and membrane tanks, it is these fibres that readily agglomerate to form braids. The mechanical integrity of these braids have been found to be unaffected by chemical treatment with sodium hypochlorite or weak acids nor are they biodegradable. The braided material blocks the membrane plates, inhibiting the coarse air scouring channel which controls the biological layer on the membrane, potentially resulting in caking on the membrane surface. The fibrous materials which braid together were also found to contain fine abrasive material in the form of fine grit. These braids rub against the membrane with the movement of the mixed liquor and air and weaken the integrity of the membrane plate. Figure 2 is a membrane plate that was removed from service which shows the effects of a blocked coarse air scouring channel and the resulting caking which can occur. The physical cleans have since been repeated on an annual basis. Whilst these enable the removal of the braiding and maintain membrane operability, the cleaning operations run for periods of up to two weeks for each of the two membrane trains, they limit plant performance, they involved managing a number of subcontractors on site undertaking highrisk work and they are a substantial cost. Hence in terms of safety, plant operability and cost, a more effective solution was needed. POSSIBLE SOLUTIONS In order to remove the fibrous suspended solids from the mixed liquor several different plant locations and screen types were considered. The ease of retrofitting the proposed solution to the existing plant with minimal operational interruptions was one of the most important factors. Improving efficiency of the existing inlet screens was explored by reducing one of the three 3mm inlet screw screens to a 2mm aperture. This was trialled for 3 months during which a minimal difference in removal efficiency was observed. Due

2 to the lack of screening in the upstream pump stations and network, it was seen that further reduction in the inlet screens below 2mm would be high risk and introduce further operational issues. Installation of secondary inlet screening had the advantage of being easily retrofit due to the plant setup and existing space available however it would require multiple, large units to enable screening of the entire inlet channel. Secondly, the smaller aperture size required to remove the fibrous material would simultaneously remove some of the organic load which in turn would require higher chemical dose rates to replace the removed carbon sources and achieve the same level of nutrient removal. Inlet screens of any form were also disadvantageous as they could not reduce the fibrous material that had already accumulated within the plant. It order to remove this existing accumulation of fibrous materials from the mixed liquor it was evident that the chosen solution needed to include screening of the mixed liquor. This could either be achieved by screening the mixed liquor upon entering the membrane tanks utilising travelling band screens or a side stream could be taken from the existing return activated sludge (RAS) line, filtered and returned. Screening the mixed liquor prior to the membrane train inlets was advantageous in that the entire mixed liquor volume would be screened. However the flow rate into the membrane tanks is equivalent to the plant inlet flow rate plus the RAS flow rate which can be 4 to 6 times the inlet flow rate. To accommodate this large flow, the mixed liquor filtration system would be relatively large, difficult to retrofit, have limited access for any maintenance or troubleshooting and would require a shutdown and partial drainage of half of the plant at a time in order to install. A side stream filtration process was considered highly favourable due to the smaller filtration unit required as a result of the reduced filtrate flow and consequently smaller required screen size. By taking the side stream from the RAS line, this also meant that the filtration unit could be installed externally to the bioreactor and membrane tanks. This was advantageous as it meant no plant operations would be interrupted throughout the installation and only a simple tap in would be required to start up the filtration process. The side stream screening process also supported the work done by Stefanski et. al. (2011) which suggested to prevent the ragging and braiding of the fibrous suspended solids, the fibrous suspended solids do not need to necessarily be completely removed, rather reduced to a lower concentration such that the degree of braiding and ragging is potentially eliminated. PILOT TRIAL To confirm the effectiveness of screening the mixed liquor and determine an appropriate screen aperture size, a 0.3m 2 filtration unit was installed on site. This unit screened mixed liquor taken directly from one of the bioreactors and was operated intermittently over a period of 3 months. The filtration unit operated by passing mixed liquor over an inclined screen using a header tank and overflow weir to ensure even distribution. As the mixed liquor flows over the inclined screen, any solids are collected whilst the permeate passes through and is collected under the unit. A pneumatic piston drives two spray arms located above and below the screen. These spray bars move up and down the screen intermittently at a predetermined rate dependant on the rate of solids deposited on the screen. This both cleans the screen and pushes the collected solids into a solids collection hopper at the end of the unit. Due to the high water content of the solids, a dewatering unit would typically be used. For the purpose of these trials the solids were allowed to drain overnight whilst the filtration unit was off which resulted in solids of 7-11%. Overall the pilot trial proved to be successful and there were no major operational issues encountered with the filtration of the mixed liquor. Throughout the period several different screens of different aperture size were trialled including a 125, 250, 500, 750 and 1000 micron. Key results from these trials are shown below in Table 1. Screen size (micron) Flux (m 3 /hr/m 2 ) Solids Removal (kg/hr) Solids Removal (kg/m 2 /hr) Improvement in filterability index minute -3% 131% 28% 27% -3% 5 minute 3% 56% 39% 13% -3% 10 minute 13% 45% 48% 2% -2% Table 1: Key Pilot Trial Results The 125 and 250 micron screens were found to be too fine and were restrictive in terms of the flux. It was also found that the two screens were removing mixed liquor as well as fibrous materials with a

3 reduction in mixed liquor of approximately 40% calculated for the 250 micron screen. Pilot results showed the 1000 micron screen was too large to effectively remove the fibrous materials. This was also noticed during operations with the downstream pump station which returned the permeate to the process continually blocking up and requiring regular cleaning to remove the buildup of fibrous materials. In terms of solids removal per hour it can be seen that the 500 micron screen performed best followed by the 750 micron screen. The 500 micron screen was also noted to have the added advantage of a considerable improvement in the filterability of the screened mixed liquor. This has the potential to reduce membrane fouling and increase the peak capacity both of which may contribute to an extension in membrane life. FULL SCALE PLANT APPLICATION A critical analysis was carried out by various TRILITY business groups to compare both the side stream filtration unit and the pre membrane tank screening option. This considered many factors such as operability, occupational health and safety, maintenance, flexibility and whole of life cost which the two different options were scored against accordingly. The side stream filtration unit proved to be the highest rating solution. This had many advantages over the pre membrane tank screening option which included having an interchangeable, sub 1000 micron screen aperture size, the ability to run intermittently without affecting plant operations and it would potentially increase the filterability of the mixed liquor over time. In terms of installation the side stream filtration unit was also superior due to its ease to retrofit to the existing plant structure without affecting plant operations. A 3m 2, 500 micron side stream filtration unit was subsequently designed and installed on site in February The filtration unit screens the mixed liquor which is taken directly from the RAS lines and has a hydraulic capacity of up to 8L/s. The unit is connected to the RAS lines such that the mixed liquor can be taken from either membrane train. The screened mixed liquor is pumped back into the same RAS line and the captured solids are dewatered and then transferred into bins for external disposal. There are future plans to incorporate this waste into the plant s existing sludge wasting lines to then be dried in the lagoons on site. Figure 3 shows the finished installation of the side stream filtration unit. RESULTS In order to measure the effectiveness of the side stream filtration unit, the current level of fibrous suspended solids in the plant needed to be determined. This was achieved using the published method by Stefanski et al (2011) which measures the concentration of coarse suspended solids (CSS500). The measurement involved passing a mixed liquor sample through a 500 micron sieve and then drying and measuring the retained solids. Mixed liquor samples were routinely taken form each bioreactor and the corresponding RAS lines for a period of 8 months to determine the base line concentration of fibrous material. On average, the CSS500 concentration was found to be 150 mg/l in bioreactor 1 and 172 mg/l in bioreactor 2. Following commissioning the filtration unit was operated intermittently whilst the equipment was optimised. During this period, large volumes of highly braided fibrous solids were observed being removed by the filtration system. After one month of operations, the average CSS500 concentration in bioreactor 1 and 2 was and mg/l respectively, equating to a decrease of 30% in bioreactor 1 and 9% in bioreactor 2. The average CSS500 concentration in the permeate of the filtration unit was calculated to be 1.7mg/L which equates to a removal efficiency of approximately 98%. Figure 4 compares two separate CSS500 analysis taken from the feed and permeate of the filtration system. In the second month of operations it was noted that the decrease in the CSS500 concentrations of both bioreactor 1 and 2 appeared disproportionate. Further investigations discovered this was due to poor compactor performance with the majority of concentrated solids being fed back to the plant at an average concentration of 1,195 mg/l. Optimisation and rectification works are currently underway to improve the performance of the compactor. Throughout the operational period of the filtration unit, there have been no adverse or negative effects on the overall performance of the plant. The mixed liquor suspended solid concentration has remained at the usual operational level with no changes to the current sludge wasting schedule. This confirms that the solids removed by the filtration unit are predominantly fibrous and no significant amount of mixed liquor is being removed. To date, there has been no improvement to the filterability of the plant; however, it is hoped once the operational refinements are complete there will be a gradual improvement similar to what was observed in the pilot trials.

4 Since the commissioning of the side stream filtration unit, TRILITY estimate that 4.5 tonne of dry material has been removed from the mixed liquor which is similar to the quantity removed during an annual cleaning operation. This reduced concentration of fibrous material will reduce the braiding or ragging potential which in turn reduces the need to remove and clean the membranes. This has the potential to save up to $100,000 every two years in maintenance costs. Additionally, as the membranes will not be subjected to the harsh operating conditions previously experienced, it is expected that the life of the membranes will also be extended. The potential cost saving of the membrane life extension is in the order of $500,000 to $1,000,000. membrane bioreactors. Ovivo Water. West Bromwich, UK. Apart from the potential plant operability and cost savings, the installation of the side stream filter will also significantly decrease the occupational health and safety risks associated with the annual physical clean of the membrane tanks. To date, these works have involved many high risk tasks including working at heights, crane operations, biological exposure and working in confined spaces. Whilst TRILITY has not had an incident in relation to this, by reducing the need for these activities, any risks are averted. CONCLUSION The installation of a side stream filtration unit has been shown to successfully remove a large quantity of fibrous suspended solids from the mixed liquor of a SMBR plant. The removal of this material has the potential to avert the need for physical cleaning in the future and possibly prolong the life of the installed membranes. Following the current optimisation program, the filter unit will be able to operate intermittently and with minimal operator intervention. Based on the observed results it is expected to see further reductions in concentration of fibrous material throughout the VHWWTP which will provide increased confidence that the membrane life and performance will not be impaired by ragging and braiding. ACKNOWLEDGEMENTS The authors would like to acknowledge the work done by Tim Grosser through the pilot trial period and SA Water for their support during testing and optimisation. They would also like to acknowledge the Victor Harbor Process Controllers for their assistance and the countless number of mixed liquor samples collected. SA Water for their support during testing and optimisation. REFERENCES Stefanski, M. Kennedy, S. Judd, S The determination and origin of fibre clogging in

5 Figure 1 : (L) Fibrous Braids on Membrane Permeate Tubes (R) Clean Membrane Permeate Tubes Figure 2: Membrane plate removed from service

6 Figure 3: Installed side stream filtration unit Figure 4: CSS500 samples before and after side stream filtration unit