Evaluation of membrane filtration and UV irradiation to control bacterial loads in recirculation aquaculture systems

Ultraviolet (UV) irradiation is commonly used to control pathogen loads in recirculation aquaculture systems (RAS), although these micro-organisms can be shielded by particles in the water, and some species tolerate very high UV doses. The objective of this study was to evaluate membrane filtration (MF) as an alternative, or complimentary, treatment to UV irradiation for pathogen control in RAS, as well as examine the operation and cost of each treatment. In a pilot-scale RAS, both MF and UV were used to treat wastewater for 30 days and water samples were collected biweekly and analysed for culturable bacteria, suspended solids, UV transmittance and other parameters. Bacterial control efficiencies were similar between both MF and UV treatments, which removed 99% of total bacteria and 98% of heterotrophic bacteria, respectively. Surface fouling was negligible for the UV while MF required biweekly cleaning to maintain operation. However, MF had the additional benefit of removing 96% of suspended solids, which resulted in increased UV transmittance. Capital and operating costs of MF were similar to UV, but only when MF treated a fraction of the wastewater compared with UV. We conclude that MF represents a potential complimentary technology to enhance UV irradiation, especially to minimise pathogens in RAS that are shielded by particles or tolerate UV.

. Recirculation of waste materials and pathogens can impart added stress on 37 fish with associated morbidity, and can increase the prevalence of infection and clinical disease (Conte, 1992; 38 Wedemeyer, 1996). Micro-screen drum filters are typically used in RAS to remove wastes greater than 50 microns, 39 but smaller particles and pathogens can often bypass these micro-screens (Patterson et al, 1999). Ozone is often used 40 in RAS to deconstruct fine particles and pathogens, but using ozone can present a health risk to fish as well as humans 41 because it is toxic at low levels (Sharrer et al., 2005;Wedemeyer, 1996). UV irradiation is typically used in RAS 42 either alone, or in combination with ozone, to inactivate pathogens, but pathogens can sometimes tolerate this 43 treatment when "fouling" or high turbidity occurs, which shields pathogens from the UV exposure (Lazarova et al., 44 1999;Liltved and Cripps, 1999;Wedemeyer, 1996). In addition, some pathogens can tolerate very high doses of UV, 45 and these proliferate within the RAS causing increased pathogen loads for the fish (Wedemeyer, 1996). For example, 46 Flavobacterium psychrophilum has been found to require four-fold higher UV doses than the recommended dose of 47 30 mJ/cm 2 (Sharrer et al., 2005) to achieve 5-log reductions of this pathogen (Hedrick et al., 2000). This bacterial 48 pathogen has been found in several aquaculture facilities of rainbow trout and in some cases can cause mortalities up 49 to 90% (Nilsen et al., 2011). Lastly, UV has been found to inactivate most, but not all bacteria in RAS, and this can 50 lead to the selection and proliferation of opportunistic pathogens that destabilizes the microbial community 51 (Attramadal et al., 2012). Therefore, ineffective control of pathogens using UV irradiation poses a potential risk to 52 fish health and the biocontrol aspects of RAS. 53 3 Membrane filtration (MF) is a process technology that physically separates solids from fluid using semi-54 permeable membranes that can be classified based on pore size: either as microfiltration, ultrafiltration, nanofiltration 55 or reverse osmosis (Madaeni, 1999;Peters, 2010). Ultrafiltration membranes have pore sizes between 0.005 to 0.02 56 µm (Madaeni, 1999;Peters, 2010;Zhou and Smith, 2002) that allow dissolved ions and water to diffuse, while 57 retaining suspended particles, protozoa, bacteria, viruses and other waste components larger than the applied pore size 58 (Guo et al., 2009). Retained wastewater is continually or periodically drained from the MF system, but particles can 59 clog and adsorb onto, or within the membrane's pore structures that results in reduced filtration rates and higher   In the 30-day study, both MF and UV treatments were very effective at achieving high removal efficiencies of 161 both total and heterotrophic bacteria (Table 2). No significant differences existed in total (Wilcoxon test; W = 50, n = 162 12, p = 0.085) or heterotrophic (W = 12, n = 5, p = 1.0) bacteria between MF and UV treatments. However, two counts 163 of total bacteria (i.e. 1300 and 1500 CFU/mL) from MF effluent collected on day four and seven were not included in 164 the analysis because they were four-fold higher than influent levels, which indicated bacterial contamination inside 165 the membrane filter. Thereafter, the membrane filter was disinfected (maintenance cleaning) every two or three days 166 afterwards to reduce contamination.

167
Mean UV intensity and transmittance were 17.4 ± 0.2 mW/cm 2 and 92.0 ± 0.5 %, which produced a mean UV 168 dose of 110.0 ± 2.0 mJ/cm 2 . The UV system did not require any cleaning because UV intensity did not drop below 169 the recommended minimum of 7.5 mW/cm 2 .

170
Continuous air-scouring, membrane relaxation cycles and maintenance cleanings were effective at reducing 171 fouling of the MF system (Figure 2). After the first maintenance cleaning, the MF system achieved a mean fouling 172 rate of 7.04 ± 1.19 x 10 4 /m/day, or a change in TMP of 0.66 ± 0.11 kPa/day based on nine filtration cycles. On day 173 0, the MF system had an initial TMP of 3.1 kPa (membrane resistance) and after 30 days of operation it had a final 174 TMP of 22.9 kPa. In addition to bacteria, MF achieved significant removal efficiencies of turbidity and total suspended 175 solids and had a significant effect on UV transmittance, pH and dissolved oxygen (Table 3).

176
The cost comparison showed that one MF element had lower capital and operating costs than UV, but less 177 wastewater would be treated (Table 4). In order to filter the same flow as UV, six MF elements would be required, 178 consequently increasing the capital and operating costs of the MF system to 3.8 times that of comparable UV system. to our study, influent UV transmittance levels were higher (i.e. 89-95%), which suggested that the UV system was 195 able to achieve maximum bacterial removal and match the efficiency of the MF system. The MF treatment may have 196 been able to achieve higher bacterial removal efficiencies than UV if challenged with higher concentrations of 197 suspended solids and bacteria, but more research is required.

9
A low rate of membrane fouling found in the present study indicates that MF has potential for long-term 215 operation in RAS as long as biweekly cleanings are performed. The highest TMP value achieved by the MF system 216 over the 30-day study (i.e. 25.2 kPa) did not reach the maximum value of 70 kPa, which requires recovery cleaning in 217 order to reduce TMP for continued operation. The membrane fouling rate may have been further reduced by applying 218 more cleaning strategies, such as permeate backwashing, while the three strategies employed in the present study were 219 sufficient. However, the MF system treated a low concentration of suspended solids, as mentioned previously, and 220 additional studies are needed to evaluate this technology in larger-scale RAS. In addition, a study that used a MBR to 221 remove nitrogen from a RAS of cod larvae (Gadus morhua) found that feeding dry feed resulted in higher membrane 222 fouling than feeding live feed . In comparison with UV, no maintenance or recovery cleaning 223 strategies were required, but again this may be due to low influent levels of suspended solids. Fouling is a larger 224 challenge for long-term operation of MF compared to UV, but further advances in membrane resistance, cleaning 225 strategies and design could reduce fouling rates and increase the long-term potential of MF in RAS.

226
New wastewater treatment technologies need to be affordable in order for their successful transition and use in 227 the aquaculture industry. Increased affordability was reflected in the present study as the cost of the single MF system 228 was cheaper and treated twice the flow rate of a similar MF system in RAS reported previously (Viadero and Noblet, 229 2002). The cost comparison showed that the advantage of MF over UV is the low capital cost of the MF elements and 230 their long lifespan, but the need for more pumps and frequent cleaning are disadvantages for MF. The capital cost of 231 the MF system with one element was lower than the UV system (Table 4). However, a MF system composed of six 232 elements are needed to treat the same flow rate as UV (i.e. 200 L/min) and this larger MF system would result in 233 greater than 1.5x the capital cost and 3.5x the operating cost. The higher cost of MF reduces its potential to completely 234 replace UV and treat a high flow rate of wastewater in RAS, but MF may be affordable as a small side-stream