Water Journal : Water Journal September 2012-1
refereed paper membrane technology water SEPTEMBER 2012 67 Given the volume of a unit (3,300L) and the flow rate through a unit (60L/s), effluent resides for approximately 50 seconds in the unit; hence, the feed and filtrate samples at any discrete time are effectively paired. The use of Method 3 for determination of the LRV will favour the selection of the 1-minute LRV as the lowest LRV during any filter run; however, this is not representative of the average water quality generated by the process over a 30-minute period. The use of Method 2, which incorporates the 1-minute samples, provides a better representation of the overall water quality, allowing for conservatism due to inclusion of the 1-minute samples with equal weighting in their determination of the LRV. Based on discussions with the regulatory authority, the 5th percentile of 2.5 log10 calculated using Method 2 was determined to be the removal efficiency of the GRWTP. This is 0.5 log10 greater than the process was initially accredited during the design phase. This result is similar to the findings of the NSF (2009), which reported an average LRV of 2.49 log10 for MS-2 while Bacillus atrophaeus, a surrogate for protozoa, recorded 6.89 log10 for the L20V membrane module in a test rig arrangement. The findings are less than those of Humbert et al. (2011), who reported an MS-2 LRV of 4-5 log10 for pressurised membrane technology at a lab scale. LRV and membrane performance Several studies have shown that pathogen removal is affected by the age of membranes -- new/mature (Jacangelo et al., 2006; Humbert et al., 2011), permeate flux and feed concentrations of challenge particulate (Farahbakhsh and Smith, 2004), transmembrane pressure (Arkhangelsky and Gitis, 2008), chemical cleaning regime (Guibert and Colling, 2011) and water quality (Jacangelo et al., 2008). Jacangelo et al. (2006) found that up to 2 log10 additional removal of viruses could be achieved by a deposition layer on the membrane surface, which would form gradually after backwashing and membrane fouling had a greater impact on removal than did deposition layer formation. Figure 4 indicates an increase in removal during filtration. Using results from Method 2, the unit LRVs were trended with respect to unit TMP and R as a function of time (Figure 5). At the beginning of the filter run, TMP tends to display a short rapid increase that becomes more stable with time. The 1-minute MS-2 sampling is within the initial rapid increase, which involves commencement of formation of the deposition layer on the membrane following introduction of feedwater to the clean membranes and, hence, displays lower removal as is evident on several occasions (Day 1: u2, u4; Day 2: all units, Day 3: u5). However, the pattern of an increase in LRV with TMP and R as a function of time was not evident for all units and a plot for the complete dataset displays a weak relationship (Figure 6). Conclusion The health regulator, based on the 5th percentile, accepted the LRV result of 2.5 log10 for the UF membrane process at the Glenelg RWTP. Conservatism was encapsulated in the testing and analysis protocol by taking into account: that low feed water turbidity and suspended solids reduced the potential for pathogen capture into flocs; test units operated at low to moderate TMP, R and FFI, which represented minimal fouling; no pre-chlorination was employed as would be in place in normal operation; and the inclusion of 1-minute LRVs, which were statistically lower than the remaining LRVs. The revision of the approval created greater flexibility and optimisation of the plant, as the chlorine contact time could be reduced from 25 to 20 mg.min/L. The membrane critical compliance limit for PDR was also revised from 4.8 to 13.4 kPa/ min (City Green Alliance, 2010b). In addition, the findings and revised approval conditions were transferable to the Aldinga Southern Urban Reuse Scheme, which incorporates the same CP L20V membrane system. This paper was originally presented at Ozwater'12 in Sydney in May. Figure 5. Trend of MS-2 LRV, Resistance (R) and Transmembrane Pressure (TMP) for UF units 2, 4, 5 and 6 across three days of testing.
Water Journal November 2012-1
Water Journal August 2012