Physical membrane damage

The turbidity and COD of MBR permeate from the control and the physical membrane damage reactors, for the first 20 minutes after the fibres were cut are illustrated in Figure 26 and Figure 27. Turbidity of the MBR permeate was not affected after cutting the first fibre. The turbidity continued to closely match the values of the control (0.2 NTU) for 10 minutes, however, after cutting the second fibre, the turbidity in the MBR permeate immediately increased to 49 NTU by 1 minute and to 360 NTU by 3 minutes. It then decreased to 4 NTU by 7 minutes suggesting that biomass had clogged and sealed the breakage. The turbidity reduced to 0.3 NTU after 9 minutes and then increased to 1.1 and 1.2 NTU at 13 minutes and 14 minutes, respectively indicating a decrease in clogging. The turbidity was reduced back to 0.4 NTU after 15 minutes, gradually reduced to 0.2 after 18 minutes, and remained stable at this level until the end of the experiment (48 hours). The results confirm that turbidity is a good performance indicator for online monitoring, providing instant indication of physical membrane damage. Per m eat e tur bi di ty (N TU ) 100 200 300 400 Control

Membrane damage - cut 1st fibre Membrane damage - cut 2nd fibre

Permeate COD concentrations of the control and the physical membrane damage experiments are presented in Figure 27. After cutting the first fibre, the COD concentration in the MBR permeate increased from 24 to 48 mg.L-1 over 3 minutes, reduced to 33 mg.L-1 by 4 minutes, and remaining stable at this level. After cutting the second fibre, the COD concentration in the permeate immediately increased to 124 mg.L-1 after 1 minute, reducing to 51 mg.L-1 after 10 minutes indicating further leakage of dissolved organic matter through the membrane breakage. Such a membrane failure is expected to impair permeate water quality (Judd et al. 2011). The COD slowly reduced to 35 mg.L-1 after 60 minutes and remained stable at this level until the end of the experiment (48 hours). The results indicate that permeate COD concentration is also a potentially useful indicator for monitoring physical membrane damage conditions. However, COD is yet to be measured as an online analytical technique.

Minutes after cutting membrane fibres

0 5 10 15 20 P er m eat e C O D (m g. L -1 ) 0 20 40 60 80 100 120 140 Control

Membrane damage - cut 1st fibre Membrane damage - cut 2nd fibre

Figure 27 - Permeate COD concentrations of the control and the physical membrane damage

experiments in the first 20 minutes after fibre cutting.

11.4

Conclusions

This is the first study that comprehensively investigated the impacts of a wide range of hazardous events on the operational parameters and key bulk water quality parameters of MBRs. The outcomes of this study will therefore facilitate greater understanding and validation of MBR processes. The main conclusions are:

• Significant reductions in COD and DOC removals were observed immediately after salinity shock, DNP shock and organic carbon shock, indicating that COD and DOC removals are effective parameters for monitoring the impacts of these hazardous events. Ammonia shock led to an immediate increase in fouling rate that was easily reversible through the relaxation period applied to the MBR 24 hours after shock, while increased fouling rates in other shock load reactors were still high 72 hours after shocks, indicating that the biomass and its impact on fouling did not recover.

• Starvation had a noticeable effect on MLVSS and MLSS concentrations, but nonetheless, the systems appeared to be resilient in terms of COD and DOC removal efficiencies. MLVSS and MLSS concentrations may be sensitive indicators of feed starvation. However, this indication may not necessarily translate into immediate performance problems.

• Turbidity and COD analyses in the physical membrane damage experiment revealed that any direct impacts were ‘self-repaired’ by the blocking of the breakage within approximately 15 minutes. The results confirmed that turbidity is a suitable performance indicator for online monitoring and able to quickly detect physical membrane damage. Permeate COD

concentration changes are also a potentially sensitive indicator for detecting physical membrane damage, but is limited by its offline nature.

• High removal rates of COD were maintained throughout the loss of biomass experiment. However, the fouling rate continued to increase considerably during the experiment, indicating unsustainable operation.

This study has identified which types of hazardous event have led to observable impacts in MBR treatment performance and permeate water quality. Future research could aim to better understand the mechanistic phenomena resulting from those hazardous events. For example, advanced microbial activity study is expected to identify specific changes in biomass characteristics during the shock loads.

12

Assessment of hazardous events on MBR:

Impact on microorganism LRV

12.1

Introduction

Membrane bioreactors (MBR) are frequently used as a barrier in water recycling schemes, where biological nutrient removal is required and plant footprint is constrained (Lesjean et al. 2011). The primary hazard in water recycling are pathogenic microorganisms originating from sewage, due to the potential for acute health effects from exposure to low dosages (van den Akker et al. 2014). In water recycling applications, a thorough understanding of pathogen removal performance and variability for each treatment barrier is imperative. Any event compromising the pathogen removal efficiency must be detected and quantified to inform appropriate corrective action (Trinh et al. 2014). The

mechanisms for pathogen removal in a MBR are size exclusion, entrainment within activated sludge flocs or membrane fouling layer and biological predation (Hai et al. 2014). Previously, theoretical simulations of hazardous events (Friedler et al. 2008) and shock loading of MBRs with domestic chemicals (Knops 2010) have been conducted, but without measurement of pathogen removal. In this study, indicator organisms FRNA bacteriophage (FRNA), Escherichia coli (EC) and total coliforms (TC), and Clostridium perfringens (CP) were chosen to represent pathogenic viruses, bacteria and protozoa. Log removal values (LRV) were quantified during operation under normal, and hazardous event conditions. Direct measurement of pathogenic species in wastewater is often not feasible due to low and highly variable concentrations and complex analysis procedures (Antony et al.

2011). As a result indicator organisms are often chosen as surrogates for pathogens. A suitable indicator organism should be chosen such that it displays correlated or more conservative removal than the target pathogen (VDoH 2013). FRNA have been investigated in several previous studies of log removal in MBR (Severn 2003, Ottoson et al. 2006, Hirani et al. 2010, Pettigrew et al. 2010, Francy et al. 2012, Hirani et al. 2012, van den Akker et al. 2014). FRNA was selected as an indicator of virus removal performance due to its small size (0.025 μm) (Antony et al. 2011) and low iso-electric point (pH 3.9) (Michen et al. 2010). With a diameter of 0.025 μm, FRNA presented a substantial challenge to removal via size exclusion by the membrane (pore diameter generally larger than 0.04 μm) and was chosen to model similarly sized enteroviruses present in wastewater. The low iso- electric point (pH 3.9) relative to the typical operating pH of MBR (7-8) (Judd 2011) reduced the likelihood of adsorption of FRNA to the membrane, as above pH 3.9 the virus particle carries a net negative charge (Antony et al. 2011). Hence, FRNA was chosen as the virus indicator given, well- documented previous use and its conservative model properties. EC and TC were chosen to represent bacterial pathogens, due to their extensive historic use as fecal contamination indicators and as challenge organisms for membrane systems. CP was selected as a surrogate for protozoa. Due to CPs ability to form spores and resist hard environments, it has been used as a surrogate for cryptosporidium in disinfection studies (Venczel et al. 1997). Depending on the strain analysed, CP spore diameters range between 0.6 – 1 .0 μm (Orsburn et al. 2008). CPs smaller size, relative to other protozoa (5 – 10 μm) (Antony et al. 2011), further supports its use as a conservative indicator in membrane challenge testing. Additionally, CP has been used as a challenge organism to represent protozoan removal in previous studies on MBR (Ottoson et al. 2006, Marti et al. 2011, van den Akker

et al. 2014).

As part of a larger investigation into operational resilience of MBRs, this paper is the third in a series that has previously assessed the impact of hazardous events on key bulk water quality parameters (Trinh et al.) and trace organic chemical removal. The aim of this study was to quantify the impact of hazardous events on removal of indicator organisms. Even under normal conditions, performances of wastewater treatment processes are inherently variable. Through the use of Monte Carlo simulation, hazardous events are evaluated with respect to process variability under normal conditions.

Benchmarking against the magnitude of normal variability provided a realistic measure and ranking of hazardous event consequence. New knowledge has be provided as a result of this study that

12.2

Experimental