Pathogen validation of small- and large-scale recycled water plants utilizing various clarification and media filtration technologies

ASR

activated sludge reactor

AWRP

advanced water recycling plant

BOD₅

5-day biochemical oxygen demand

CFU

colony forming units

Ct

contact time

DAF

dissolved air flotation

DAFF

dissolved air flotation filtration

μDALY

micro disability-adjusted life year

HBT

health-based targets

LRV

log₁₀ reduction value

MAR

managed aquifer recharge

PCT

physical–chemical treatment

PFU

plaque forming units

PMF

pressurized media filtration

RWTP

recycled water treatment plant

SS

suspended solids

TKN

total Kjeldahl nitrogen

UV

ultraviolet

WWTP

wastewater treatment plant

Reduced access to fresh water, driven by population growth, increased industrial and agricultural use, and climate change, has led to a rise in wastewater recycling as communities seek secure, ‘climate-independent’ water supplies (Short et al. 2022). As a result, recycled water has become a key component of diverse water supply portfolios for many water utilities in countries like Australia, where it is used for various purposes, including residential use via dual reticulation systems, and in agriculture for the irrigation of crops (Short et al. 2022).

Recycled wastewater poses increased risks to human health due to the types of end-uses and exposure pathways. Consequently, recycled water guidelines, such as those in Australia, adopt a quantitative risk-based management approach to protect public health against pathogens through the setting of health-based targets (HBTs) (NRMMC et al. 2006). Microbial HBTs are treatment targets, expressed as log₁₀ reduction values (LRVs) for reference organisms that represent the major sewage pathogen groups (bacteria, viruses and protozoa) (Petterson & Ashbolt 2016). A key requirement for utilities and regulators is to ensure that schemes meet HBTs, which is also dependent on the end-use exposure routes, but in all cases to meet a health benchmark of <1 μDALY/y (NRMMC et al. 2006).

To meet HBTs, wastewater recycling systems often require the use of multiple treatment barriers, and one such example lies in the selection of appropriate filtration methods. Media filters, in particular, are often used to maximize the disinfection efficiency of downstream processes (e.g. chlorine and ultraviolet (UV)) and to act as a physical barrier for the removal of enteric pathogens (Gray 2014). In practice, this can be achieved through either pressurized or gravity media filtration. Pressurized filters can operate and handle higher flow rates, have a high filtration efficiency, and therefore are more compact in design, making them suited to small-scale reuse schemes with low space requirements. Conversely, gravity filters require a larger footprint and are suited for large-scale schemes due to their ability to handle high flow volume.

When selecting a filter type, a key consideration in the decision-making process is the ability of media filters to remove pathogens to a level that can assist a scheme in achieving its pathogen HBT. Reported LRVs from media filtration vary between 0 and 1.0 log₁₀ (for bacteria), 1.0 and 4.0 log₁₀ (phage viruses) and 1.0 and 3.0 log₁₀ (for protozoa) (NRMMC et al. 2006), and much of the published data regarding pathogen removal through media filters has been derived from their application in various drinking water treatment plants (Mazounie et al. 2000; Hijnen et al. 2004; Boudaud et al. 2012; Asami et al. 2016; Kato et al. 2018; Shirasaki et al. 2018; Canh et al. 2019; Nilsen et al. 2019; Sylvestre et al. 2021; Pang et al. 2022; Shirakawa et al. 2022). Moreover, research into pathogen removal performance of filtration systems treating recycled wastewater has largely focused on membrane-based technologies (Lonigro et al. 2006; Pype et al. 2016; Reeve et al. 2016; Carvajal et al. 2017; Lee et al. 2017; Branch et al. 2021). Subsequently, validation protocols have been developed (such as in the US and Australia) for membrane bioreactors, reverse osmosis and nanofiltration, providing frameworks for measuring pathogen removal from these systems.

In contrast to membrane-based filtration technologies, comprehensive data on pathogen removal through different types of media filters applied to wastewater recycling, especially those operating at full scale is limited (Fu et al. 2010b; Rodriguez-Manzano et al. 2012; Schmitz et al. 2024). This lack of full-scale data and uncertainty has meant that health regulators may assign conservative default pathogen LRVs that can underestimate the true performance of media filters which adds significantly to operational costs. Consequently, this presents challenges for planners in selecting the appropriate technology to meet HBTs and achieve fit-for-purpose water recycling.

To address this gap, the present study undertook an investigation into the pathogen LRV performance of two different types of media filtration technologies from two full-scale wastewater recycling schemes that were situated within the same wastewater treatment plant (WWTP). The first was a pressurized media filter (PMF), from a small 17,000 m³/d capacity water recycling scheme designed with large product water storages, enabling continuous operation irrespective of seasonal variations in demand. The second filter type was a gravity filter in the form of a dissolved air flotation filtration (DAFF) process from a large 120,000 m³/d capacity water recycling scheme, designed without product water storages, that operated intermittently to accommodate seasonal changes in farmer demand. Validation of protozoa and virus removal was estimated by dosing model organisms yeast (Saccharomyces cerevisiae) and MS2 bacteriophage at high densities to demonstrate log₁₀ reduction potential. Log₁₀ removal through the water clarification processes which lie upstream of the filtration processes were also assessed given these processes directly impact the effectiveness of the filtration systems. This included comparing LRVs through lamella sedimentation and flotation processes.

This study therefore presented a unique opportunity to validate and compare the performance of two reuse schemes, which are situated within the same WWTP site, operated at different scales, and employing different media filtration technologies, while achieving a common objective of providing recycled water for unrestricted use on commercial food crops, non-food crops and municipal irrigation.

Additionally, this study assessed and compared the energy and chemical consumption demands associated with the different treatment strategies. The capability to evaluate energy and chemical usage data alongside pathogen removal performance from full-scale water reuse schemes, which vary in their technological approaches, offers an opportunity for comparing and benchmarking schemes. The information presented in this paper, therefore, offers significant practical insights into the technological and economic aspects, aiding decision-making in the design of reuse schemes that are fit for purpose.

The RWTP is one of the largest schemes in Australia providing reclaimed wastewater to approximately 360 customers and is primarily used for horticulture. The RWTP draws its source water from large stabilization lagoons (348 ha), which operate at a minimum 16-day rolling average retention time for the removal of enteric protozoa. The plant has a large treatment capacity of 120,000 m³/d and experiences high seasonal variations in demand. The source water can be challenging to treat during the irrigation season, as higher summer demand in reclaimed water coincides with periods of high suspended solids (SS) caused by blooms of phytoplankton within the lagoons. A DAFF process is used to treat this water which is comprised of (1) flash mixing chambers that are dosed with a coagulant (aluminium sulphate), and then polymer (Magnafloc LT510); (2) clarification using 12 parallel dissolved air flotation (DAF) and (3) subsequently filtration, located on the bottom of the DAF unit, through dual media, sand (1.3–3.0 mm in diameter) and anthracite (3.0–6.0 mm in diameter), on a gravel layer (6.0–12 mm in diameter). The common filtered water is chlorinated to achieve a chlorination contact time (Ct) required to achieve a 1 − log₁₀ reduction of viruses. While the stabilization lagoons provide challenging water for the DAFF process to treat which exerts a high chemical (coagulant) demand, the lagoons serve as an important low-cost barrier for the removal of protozoan pathogens.

In contrast to the RWTP, the AWRP has a lower treatment capacity (17,000 m³/d), is compact in size, but has large water storage capacity for periods of low demand (winter), which is comprised of 400,000 m³ capacity earth banks and a managed aquifer recharge (MAR) scheme that is licenced by the Environmental Protection Authority to store 4 GL/y. The AWRP receives secondary treated effluent from the secondary clarifiers, which has a lower solids content and is treated using compact treatment technology comprising physical–chemical treatment (PCT), including coagulation (aluminium sulphate), flocculation (polymer) and clarification (lamella sedimentation). The clarified water is then filtered via a set of (six in total) PMF units (Industrias Metalicas Imetal, S.A.) comprised of sand and anthracite. Unlike the RWTP, the AWRP does not benefit from protozoa removal offered by the stabilization lagoons, and as such, energy-intensive UV disinfection (LBX1500E, WEDEC) is required post-filtration to meet the HBT objectives for protozoa. The water is then further disinfected with sodium hypochlorite (13% solution) to achieve a chlorination Ct required to achieve a 4 − log₁₀ reduction of viruses.

Both water recycling schemes utilize multiple barriers to fulfil virus, bacteria and protozoan HBT required for unrestricted irrigation. For the AWRP, the treatment barriers include the activated sludge process, PCT and PMF, UV and chlorine disinfection. Treatment barriers utilized by the RWTP include activated sludge treatment, stabilization lagoon treatment, the DAFF process and chlorine disinfection.

All validation challenge studies were undertaken under each plant's maximum steady flow rate (Table 1) over the course of one day. The coagulation and flocculation set points, and surrogate dosing parameters for each challenge test are highlighted in Table 1. For all validations, backwash frequency occurred in-line with the normal operation.

Validation was based on characterizing the removal of protozoa and enteric viruses using surrogate organisms which were baker's yeast (S. cerevisiae) and MS2 bacteriophage. Both organisms are accepted as model organisms to assess the effectiveness of filtration systems (Davies et al. 2008; Victorian Department of Health 2013; Nilsen et al. 2019; Sharaf et al. 2020; Water Research Australia 2024). This is because the size of S. cerevisiae and MS2 bacteriophage are equal to or less than protozoan pathogens (e.g. Cryptosporidium spp. and Giardia lamblia) and human enteric viruses, respectively (Nilsen et al. 2019; Sharaf et al. 2020). They can also be dosed at high densities to demonstrate high magnitude log₁₀ reduction. The characterization of bacteria removal was of lesser significance due to the utilization of chlorine disinfection in both schemes, which effectively satisfied the majority of the HBT pertaining to bacterial pathogens. Validation of the RWTP's coagulation, flocculation and flotation processes and the AWRP's coagulation, flocculation and sedimentation processes which lie upstream of the filtration processes were also assessed given these processes directly impact the effectiveness of the filtration systems.

MS2 bacteriophage was cultivated as described in Reeve et al. (2016) to provide a highly concentrated stock culture of MS2 (10¹¹–10¹⁴ PFU/mL). Enumeration of the MS2 bacteriophage was undertaken within 24 h of sampling and analysed by NATA certified laboratory (AWQC, South Australia) using the agar overlay technique. Yeast (S. cerevisiae) was analysed based on the membrane filter technique for yeast and fungi by APHA (1998). Briefly, dilutions were performed in sterile phosphate-buffered saline. The diluted samples (100 mL volume) were filtered through a 0.45-μm filter and then placed on Dichloran Rose Bengal Chloramphenicol Agar and incubated for 5 days at 20 °C.

A schematic representation of the RWTP and AWRP dosing set-up is shown in Supplementary Data Figure 1. For the MS2 phage validation, the batching tank was filled with dechlorinated potable water and fitted with a mixing impeller (20–30 rpm, Cole-Parmer, US). For the yeast validation, a commercial quantity (1,000 L) of liquid creamed baker's yeast (S. cerevisiae) was sourced (AB Mauri, NSW, Australia) and transported in an Intermediate Bulk Container which acted as the batching tank. No mixing of the yeast stock was required given that large CO₂ production by the yeast maintained the stock in suspension. The yeast stock was pumped either using peristaltic dosing pump for the RWTP or dosing pump for AWRP (DDA, Grundfos, South Australia) which continuously dosed the yeast into the raw water intake to achieve the target challenge doses seen in Table 3. Dosing commenced 90 min prior to the challenge tests, to ensure that each plant was primed throughout with an even density of MS2 phage or yeast.

For all validation exercises, dosing tank samples were taken (n = 3) to ensure the stock concentration of the surrogate was maintained throughout the testing timeframe. For each validation exercise, 20 influent water samples were taken prior to any chemical addition. In addition, 20 effluent samples were taken post-filtration, prior to any form of disinfection (UV or chlorination). For the AWRP validations, a further 10 samples were taken after the clarifier to understand the relative contribution of the PCT (coagulation, flocculation and sedimentation) and PMF processes to the overall systems LRV performance. During the DAFF validations, 5–20 samples were collected from above the filter bed to understand the relative contribution of the coagulation, flocculation and flotation process versus filtration in the removal of the surrogates.

Total algae count (RWTP only), biological chemical demand, pH, total phosphorus, SS and total Kjeldahl nitrogen (TKN) were analysed on influent and effluent samples during each validation exercise by a National Association of Testing Authorities certified laboratory (Australian Water Quality Centre, South Australia), using Standard Methods for the Examination of Water and Wastewater (APHA 1998). All samples were collected and stored at 4 °C until being processed.

The Shapiro–Wilk test was conducted to assess data normality. Non-parametric Spearman's correlation was used to identify statistical relationships between microbial LRV performance for both DAFF and PMF systems and effluent turbidity. To determine statistical differences in the effectiveness of yeast and MS2 removal between the treatment technologies, either a Wilcoxon rank-sum test or unpaired t-test was conducted, depending on whether the data were determined to be non-parametric or parametric, respectively. Statistical significance was accepted at the p < 0.05 level, with all analyses performed using SPSS (IBM, version 28.0.1.1 (14)).

The influent and effluent water quality measured at both the RWTP and the AWRP are presented in Table 2. The AWRP and RWTP source waters were extracted from different locations along the WWTP, and for this reason, there were notable differences in 5-day biochemical oxygen demand (BOD₅) and SS concentrations and pH at each recycled water plant. The AWRP plant, which directly receives secondary effluent, typically has an SS concentration of 12.7 ± 12.8 mg/L (measured from June 2021 to June 2022; n = 63). During the AWRP validation exercises, the influent SS concentrations ranged between 11 and 13 mg/L, effectively representing typical SS concentrations. Influent BOD₅ was below 10 mg/L and the pH was neutral which was optimal for alum coagulation. In comparison, the average SS concentration within the RWTP feed water was 79.1 ± 56.7 mg/L (measured from June 2021 to June 2022; n = 66). The RWTP sources its water from stabilization lagoons, and as a result, the influent pH, SS and BOD₅ concentrations are highly variable due to seasonal changes in phytoplankton abundances, which reached densities as high as 6.7 log₁₀/mL during summer and thereby requires high doses of coagulant. Phytoplankton blooms also resulted in elevated water pH (>9.7) during the day, posing further challenges to the coagulation process. For these reasons, pathogen validation of the RWTP was performed during summer, under high feed water SS concentrations of 197 and 96 mg/L for the yeast and MS2 bacteriophage challenge tests, respectively. In terms of the filtered water turbidity, the AWRP produced more consistent and superior effluent quality when compared with the RWTP; however, this did not translate into improved virus removal performance, which is detailed below.

The influent density of MS2 bacteriophage during the AWRP (4.9 ± 0.1 log₁₀ PFU/mL) and the RWTP (4.4 ± 0.3 log₁₀ PFU/mL) validation exercises were both consistent for the duration of the challenge tests (Table 3). The results obtained from the AWRP showed that the PCT process, which is made up of coagulation, flocculation and sedimentation, contributed the most to the MS2 bacteriophage removal, achieving a mean reduction of 0.72 (±0.22) log₁₀ compared with the PMF which achieved a mean reduction of 0.30 (±0.27) log₁₀ (Table 3). Both the PCT and the PMF maintained stable removal performance for the duration of the challenge test, providing a total combined virus reduction of 0.98 (±0.37) log₁₀.

The removal of MS2 bacteriophage through the RWTP coagulation, flocculation and flotation steps was comparable to the AWRP PCT process, which achieved a mean reduction of 0.75 (±0.12) log₁₀. When comparing the performance of the AWRP coagulation, flocculation and sedimentation processes with that of the RWTP coagulation, flocculation and flotation steps, no statistical difference in virus LRV performance was observed (Wilcoxon rank-sum test; p>0.05). This highlights that use of high-rate sedimentation or DAF as processes to clarify water provided similar results. These values are lower than what has been reported in laboratory studies. For example, jar tests conducted by Shin & Sobsey (2015) demonstrated 2.0 log₁₀ removal of MS2 phage via alum coagulation, flocculation and sedimentation. A pilot study by Koivunen & Heinonen-Tanski (2008) demonstrated that the removal of F-specific RNA (FRNA) coliphages through a tertiary DAF process ranged from 0.53 to 1.51 log₁₀, with efficiency dependent on a range of operation conditions. They also noted that LRV increased to 2.7 log₁₀ under high coagulant doses for treating primary effluent.

The RWTP's DAFF filters contributed the most to virus removal, with a mean reduction of 2.35 (±0.91) log₁₀ and displayed superior virus removal when compared with the AWRP PMF (Table 3). Together, the RWTP coagulation, flocculation and DAFF steps provided a higher total reduction of 2.90 (±0.64) log₁₀, which was 1.9 log₁₀ greater than that achieved by the AWRP PCT and PMF combined. This highlights a clear difference in filter LRV performance between the two plants. Mean virus LRV achieved by the DAFF process was at high-end for what has been reported by filtration systems and comparable to what has been achieved by ultrafiltration and microfiltration membranes processes (Reeve et al. 2016, 2017; Nasir et al. 2022) and some membrane bioreactors (Branch et al. 2021). It was unclear why virus removal through the DAF filter was significantly higher than that through the PMF; however, it was possible that the operational conditions, including a (i) lower operating pressure led to a longer filter contact time and (ii) high alum dose and elevated solid (algal floc) loading which may have been more conducive to phage clumping and attachment to larger particles, or straining since divalent cations promote MS2 adsorption (Farrah 1982).

The enhanced MS2 removal seen through the DAFF plant may also be attributed to the limitations of using MS2 phage. MS2 phage is well established as a surrogate for enteric viruses in performance testing of filtration processes since it overcomes well-known challenges of using actual enteric pathogens (see Water Research Australia 2024). Consequently, guidelines and validation protocols recommended seeded MS2 phage as surrogate organisms to validate virus removal in media and membrane filtration systems and water purifiers (USEPA 2005; Victorian Department of Health 2013; Seacord et al. 2023; Water Research Australia 2024). Nevertheless, there are recognized limitations in the use of MS2 that should be considered, including varying inactivation or removal in some systems compared with other viruses, due to differences in surface properties. Previous studies have demonstrated that exposing viral suspensions to air–water interfaces (AWIs), via shaking, bubbling, or aerosolization, leads to virus inactivation (Adams 1948; Thompson & Yates 1999). MS2 is a hydrophobic virus (Shields & Farrah 2002) that is more attracted to AWI, and hence, their inactivation can be enhanced in systems where AWI is continuously generated (Thompson & Yates 1999). In the context of this study, the inactivation of MS2 at AWIs may be more pronounced within the DAFF process since it relies on the generation of microbubbles for floc flotation, and hence, the use of MS2 may overestimate LRV performance of hydrophilic viruses. The high ionic strength of a solution is also a factor that can strengthen AWI attraction and enhance MS2 inactivation (Thompson & Yates 1999). In unsaturated soils, it has been suggested that hydrophobic viruses are less likely to be transported because of retention at air–soil–water interfaces (Gerba 1984; Sinclair et al. 2012). Viruses with lower hydrophobicity travel greater distances due to fewer interactions with interfaces (Sinclair et al. 2012). Hence, how representative MS2 is as a surrogate to estimate the removal of enteric viruses (which have varying degrees amphipathicity) through rapid filtration systems requires further investigation and may require a re-evaluation of surrogate selection for DAFF systems.

In this study, baker's yeast was used as a model organism for estimating Cryptosporidium oocysts removal at full scale, as demonstrated in previous studies (Chung et al. 2004; Davies et al. 2008; Sharaf et al. 2020). The concentrations of yeast maintained within the AWRP and RWTP feed water remained highly consistent measuring 6.8 (±0.1) and 6.9 (±0.06) log₁₀ CFU/100 mL, respectively, for the duration of the trials (Table 3). At the AWRP site, the dominant yeast removal process was the PMF, demonstrating a mean reduction of 3.5 (±0.11) log₁₀. In contrast, the PCT process alone achieved a lower reduction of 1.0 (±0.12) log₁₀.

Yeast removal through the RWTP DAFF process showed that the relative contribution of coagulation, flocculation, flotation and filtration to yeast LRV was more evenly balanced. The coagulation, flocculation and flotation processes achieved a mean reduction of 1.78 (±0.41) log₁₀ compared with 2.11 (±1.26) log₁₀ via filtration.

When comparing yeast LRV performance through the AWRP and RWTP, the AWRP PMF outperformed the RWTP DAFF filters by 1.4 log₁₀. Conversely, the RWTP coagulation, flocculation and flotation processes were able to remove 0.78 log₁₀ more yeast compared with the AWRP coagulation, flocculation and sedimentation processes. Observations seen here at full scale with yeast have also been confirmed at bench scale using Cryptosporidium parvum oocysts, where Plummer et al. (1995) demonstrated that DAF was a superior clarification process for the removal of oocysts (0.38–3.7 log₁₀ reduction) compared with sedimentation (0.00–0.81 log₁₀ reduction). Further pilot plant experiments have demonstrated that DAF is more efficient than plate sedimentation in eliminating Giardia cysts and Cryptosporidium oocysts (Edzwald 2010). Additionally, various studies have highlighted that DAF outperforms sedimentation notably in algae removal, possibly due to the microbubbles' capability to float small and low density particles (Henderson et al. 2008; Edzwald 2010). This contrasts with sedimentation, where particle settling relies solely on their density and size.

Despite variations in the relative contributions of coagulation and filtration treatments implemented at both sites to yeast removal, the overall LRV performance of the AWRP and RWTP were comparable (Table 3). The mean LRV measured at the AWRP (4.57 ± 0.14 log₁₀) was marginally higher than the RWTP (3.80 ± 1.06 log₁₀). Although yeast removals were similar, mean LRVs were statistically different (unpaired t-test; p < 0.05) and a closer analysis of the data uncovered notable variability in the RWTP performance. The 5th percentile LRV for RWTP was 2.2 log₁₀, contrasting with the AWRP, which achieved a more consistent 4.4 log₁₀ (Table 3). Further discussion on this variability is provided later. LRVs achieved here in our study using yeast exceed values reported elsewhere. For example, Fu et al. (2010a) measured the removal of protozoa through a full-scale wastewater reuse plant and found that LRVs of Cryptosporidium and Giardia were 0.77 and 0.73 log₁₀ for conventional flocculation and sedimentation, and 0.92 and 0.89 log₁₀ for sand filtration, respectively. Conversely, data reviewed by Gray (2014) reported that LRVs for protozoa in the order of 2.0–3.0 log₁₀ were possible when using a combination of coagulation, sedimentation and media filtration with good operation management.

Characterizing the variability in treatment performance is an important aspect to consider when evaluating health risks associated with recycled water (Van den Akker et al. 2014). For wastewater recycling schemes, there is a particular interest in understanding how treatment processes perform at the margins. Consequently, individual treatment barriers are typically accredited based on their validated 5th percentile LRV as a conservative performance indicator. Furthermore, there is a need to understand how variability in LRVs relates to process parameters (e.g. filtered water turbidity) that are measured online and serve as a reliable predictor of pathogen removal for continuous assessment of treatment performance and improved safe operation (Adelman et al. 2024).

Results show that the RWTP DAFF process displayed more variability in yeast and MS2 phage removal, which was mostly attributed to the filter's performance, given that LRV attributed to coagulation, flocculation and flotation was comparatively stable during the validation exercises (Table 3). This variability reflects the challenging nature of the source water, characterized by elevated algal solids and more frequent filter backwashing, as well as suboptimal pH levels (>9.7) for coagulation. A time series plot that shows the split between LRV from coagulation, flocculation and flotation versus LRV from filtration for yeast is presented in Supplementary Data Figure 2. These results showed that a drop in the DAFF filter LRV performance coincided with a sharp increase in filter effluent turbidity (from 0.4 to 1.5 NTU), which was caused by 2 out of 12 filter cells backwashing in close succession to each other. A strong negative correlation was found between effluent turbidity and yeast LRV (Spearman's ρ = −0.763; p < 0.001). It is also important to note that the yeast LRV values from the first three timepoints in Supplementary Data Figure 2 may represent outliers and were omitted from analysis since these high values indicate that the RWTP had not been fully primed with yeast at the time of sampling the effluent, which overestimated removal.

Most of the variability seen in MS2 phage removal by the DAFF process was also attributed to the filtration step (Supplementary Data Figure 2). There was however no correlation between filtered water turbidity and MS2 LRV (Spearman's ρ = −0.143; p > 0.05). This may be due to the narrow range of the filtered water turbidity (Table 2) at the time of validation, but also because effluent turbidity measurements can be a poor predictor of virus removal from filtration systems (Shirasaki et al. 2018). This has also been observed in membrane systems, where turbidity did not correlate with the levels of small particles (Adelman et al. 2024).

In contrast to the RWTP, the AWRP was more stable in terms of yeast and MS2 phage LRV performance and filtered water turbidity. The 5th and 95th percentile LRVs ranged between 4.4 and 4.8 log₁₀ for yeast and 0.38 and 1.20 log₁₀ for MS2 phage, while effluent turbidity ranged between 0.13 and 0.20 and 0.11 and 0.14 NTU during the yeast and MS2 phage validation exercise, respectively. Given the stable nature of the AWRP performance, no correlation between turbidity and yeast LRV (Spearman's ρ = 0.067; p > 0.05) or virus LRV (Spearman's ρ = 0.352; p > 0.05) was established.

Energy and chemical usage of treatment barriers constitute a significant portion of the operational costs associated with running water reuse schemes. Their assessment alongside pathogen LRV performance provides an opportunity for comparing and benchmarking schemes, aiding in the pursuit of developing recycled water plants that are fit for purpose.

In the context of this study, aluminium sulphate, which was required by the DAFF and PCT processes for the purpose of coagulating SS prior to filtration, and chlorine, which was required for disinfection, were the dominant chemicals used. Electricity usage of recycled water plants can also be significant, but it is highly dependent on the type and scale of the treatment technologies applied (Plappally 2012; Short et al. 2015). To compare electricity, aluminium sulphate and chlorine usage between the two reuse schemes, these parameters were normalized against the volumetric flow (Table 4). Electricity submetering of individual barriers was not available, so the data presented here represent whole-of-plant electricity usage. Virus and protozoa LRVs for each barrier are also provided for comparison in Table 5. For the DAFF and AWRP (PCT plus PMF) processes, the 5th percentile LRV values generated from this study are used in Table 5, given that barriers are often accredited based on their validated 5th percentile LRV as a conservative performance indicator.

Results showed that when considering treatment options pertaining to LRV performance for protozoa and virus, clear trade-offs exist between energy and chemical usage. For example, the large 120,000 m³/d RWTP scheme utilized 2.6–2.9-fold less energy than compared with the smaller 17,000 m³/d AWRP which was most likely due to economies of scale effects and the use of extensive tertiary lagoon system (348 ha) as an alternative to UV disinfection, which alone can consume 0.026–0.30 kWh/m³ (Short et al. 2015). While the RWTP was more energy-efficient, its DAFF process required 2.6-fold more aluminium sulphate than the AWRP during the peak irrigation season to eliminate high concentrations of SS in the feed water. This increase was attributed to summer blooms of phytoplankton within the tertiary lagoons. For the RWTP, the combination of the lagoon and DAFF processes provided superior virus LRV (3.0 log₁₀), but lower protozoa removal (4.0 log₁₀) when compared with the equivalent barriers offered by the AWRP (i.e. PCT, PMF and UV). Conversely, the AWRP, which consumed more energy and significantly less aluminium sulphate (per volumetric flow) than the RWTP, achieved superior protozoa removal through its PCT, PMF and UV barriers (8.0 log₁₀), while virus removal was only 1.5 log₁₀. Consequently, the AWRP scheme was required to operate at higher chlorine Ct to compensate for a shortfall in virus LRV, resulting in higher chlorine consumption.

Both reuse schemes displayed a decrease in energy efficiency during periods of low flow (winter) highlighting an economies of scale effect. However, this effect was most pronounced in the smaller AWRP, where energy requirements (kWh/m³) increased by 38% during winter. This highlights the benefit of making full use of the treated water storage options during periods of low demand which would enable operating the system at higher production flow rates to exploit economies of scale effects. However, trade-offs with cost associated with water storage and extraction might need to be considered.

Water clarification and filtration processes are crucial for pathogen removal in wastewater recycling, but a significant data gap on their full-scale efficacy for reuse applications hinders planners in selecting the right technology to meet HBTs and ensure fit-for-purpose water recycling. This study evaluated the pathogen removal performance of two full-scale filtration technologies, including PMF and gravity filtration in the form of DAFF, as well as the preceding clarification processes, by measuring reductions in model organisms.

Results demonstrated that both reuse plants were capable of achieving high magnitude log₁₀ reduction of yeast. LRVs were similar, measuring 3.80 (±1.06) through DAFF and 4.57 (±0.14) through coagulation, sedimentation and PMF. Filtration was the dominant process for removing yeast in both systems; however, the use of DAF was a superior clarification process for the removal of yeast compared with lamella sedimentation. Notably, this study demonstrated that the DAFF process was a more effective barrier for viruses compared with the combined compact lamella clarification and PMF process, achieving LRVs comparable to some membrane-based filtration systems. Higher MS2 removals may be an artefact of inactivation from AWIs, which requires further investigation.

Analysis of energy and chemical data revealed that larger capacity DAFF plant demonstrated economies of scale by being more energy-efficient in meeting its HBT requirements compared with the more compact lamella sedimentation and PMF processes. However, this advantage was offset by a higher utilization of chemicals for coagulation. The characterization of energy and chemical data alongside pathogen LRV is an important step to enable the benchmarking of schemes and provide important insights that can assist the water industry to better design ‘fit-for-purpose’ recycled water systems.

The authors thank the Suez RWTP and AWRP plant operators for their assistance.

This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sector.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.