MBR and GAC filtration followed by UV disinfection – implications for wastewater reuse at full scale

Shifting precipitation patterns due to climate change, combined with population growth and urbanization, have increased water stress and the need for alternatives to dwindling freshwater resources (Ungureanu et al. 2020). Parallel with the implementation of wastewater treatment aimed to remove organic micropollutants, for example in Sweden, these trends have generated recent interest in wastewater reuse also in areas where it has not previously been common practice (Takman et al. 2023).

Recycled wastewater has several potential uses including agricultural irrigation (Ofori et al. 2021) and direct or indirect drinking water production (Drewes & Horstmeyer 2016). Irrigation and potable reuse can reduce the demands on limited surface water and groundwater resources, and irrigation with reused water has potentially positive effects on soil and plants, due to its nutrient content (Vergine et al. 2017). Yet, domestic wastewater reuse can pose risks from chemical pollutants, such as organic micropollutants (Fatta-Kassinos et al. 2011; Verlicchi et al. 2023) or salts (Muyen et al. 2011) and pathogens, including bacteria, parasites, and viruses (Jaramillo & Restrepo 2017).

One option for removing organic micropollutants and other chemical contaminants from water is granular activated carbon (GAC) filtration (Frank et al. 2015). Full-scale GAC filters for the removal of organic micropollutants from wastewater have been implemented in, for example, Sweden (Takman et al. 2023; Svahn & Borg 2024) and Germany (Neef et al. 2022) and can be applied for the purpose of wastewater reuse (Reungoat et al. 2010; Nahrstedt et al. 2020; Hogard et al. 2021). In addition to adsorbing chemical pollutants, GAC filters can alter the microbial water quality due to biofilm growth on the granules (Weber et al. 1978; Gibert et al. 2013; Kantor et al. 2019) and increase effluent bacteria concentrations (Wilcox et al. 1983; Miller et al. 2020).

GAC filters generally do not remove microbial contaminants (Drewes & Horstmeyer 2016), but a small decrease in the concentrations of certain indicator bacteria has been observed after GAC filtration (El-Zanfaly et al. 1998; Hijnen et al. 2010; Spit et al. 2022). The additional removal of pathogens requires different disinfection processes, for example, ozonation (Facile et al. 2000; Xu et al. 2002), chlorination (Sun et al. 2009), or use of ultraviolet (UV) radiation (Hassen et al. 2000; Liberti et al. 2003). Compared with chlorination and ozonation, UV radiation does not generate any harmful disinfection byproducts, nor does it, as chlorination, result in a residual disinfection that could prevent regrowth in the distribution system (Wang et al. 2021).

The disinfection effect from UV treatment varies between microorganisms. A high log removal of the indicator bacteria Escherichia coli and other coliform bacteria can in general be achieved (Sommer et al. 2000; Mezzanotte et al. 2007), whereas higher UV fluences are required to attain the same removal of Clostridium perfringens (Hijnen et al. 2006; Carabias et al. 2023) and yeast (Spotte & Buck 1981). Enterococci can be inactivated to the same extent as E. coli and other fecal coliforms (Jacangelo et al. 2003), but lower inactivation has also been reported (Locas et al. 2008). Further, UV radiation can inactivate pathogens that can be persistent in chlorination (Hijnen et al. 2006). It is important to note, however, that UV does not necessarily remove the entire microbial population, but can shift the type of bacteria present in the community (Pullerits et al. 2020).

UV radiation damages the DNA of the microorganisms – an effect that is influenced by the absorbance of the solution being exposed, and thus by the suspended solids content of the water (Qualls et al. 1983; Carré et al. 2018). This means that the effectiveness of the UV treatment is dependent on the upstream treatment processes and the extent to which they remove color and suspended solids (Venditto et al. 2022). The combination of GAC and UV is an interesting approach for treating wastewater for reuse (Ponce-Robles et al. 2020), since it comprises both a chemical and a microbial barrier, but does not generate potentially toxic disinfection byproducts. Further, unlike reverse osmosis, which is also frequently applied for wastewater reuse, this combination does not result in a concentrate that can be costly to handle (Kehrein et al. 2021).

In the Scania region in southern Sweden, there are currently two wastewater treatment plants (WWTPs) that are equipped with GAC filtration, with disparate upstream wastewater treatments: S:t Olof WWTP with a conventional activated sludge (CAS) process and subsequent sand filtration, and Kivik WWTP with an membrane bioreactor (MBR). Both processes are common for wastewater treatment, but result in considerable differences in the effluent water quality, especially regarding particles and bacteria (Drewes & Horstmeyer 2016), potentially affecting the downstream GAC treatment. The biofilm development in full-scale GAC filters continues for long time periods after commissioning of the filters (Takman et al. 2023), necessitating full-scale studies of GAC filters with mature biofilm to examine their effects on the microbial water quality.

UV disinfection is planned for incorporation downstream of the MBR + GAC process in Kivik, with its commissioning expected in 2024, and the treated water is planned to be reused for irrigation of, for example, recreational spaces. The effluent from Kivik WWTP is currently of a high chemical and microbial water quality, that in many ways is similar to that of source waters used for drinking water production (Takman et al. 2023). The aim of this study was to expand on our knowledge of and experience with these treatment processes, by comparing the effects of UV disinfection of the S:t Olof and Kivik GAC filter effluents, and assessing the potential of the treated effluents for reuse for irrigation or drinking water production.

Microbial water quality through two different full-scale wastewater treatment processes that include GAC filtration was examined and UV disinfection of these effluents was performed in laboratory experiments. Samples were collected on three occasions (June 12 and 26, and October 2, 2023). Based on the results of the UV experiments and microbial analysis, one sample was selected for extensive chemical analysis of over 100 parameters in the Swedish drinking water criteria (LIVSFS 2022:12).

UV disinfection was conducted at the laboratory scale on effluent samples from both WWTPs within three days after sampling. Three UV fluences (approximately 200, 400, and 700 J/m²; Table 1) were tested using an Aquada 1 UV reactor (Wedeco GmbH, Herford, Germany, length 470 mm and diameter 70 mm) with a monochromatic, low-pressure mercury lamp that emits UV radiation at 254 nm. The range in UV fluences was selected because 400 J/m² is a conventional UV fluence at Swedish drinking water treatment plants (Saguti et al. 2022).

Microbial analysis (E. coli, total coliform bacteria, and total cell concentration (TCC)) was conducted in-house, with select analyses conducted by an external accredited service (Eurofins, heterotrophic plate count (HPC), slow-growing bacteria, C. perfringens, intestinal enterococci, yeast, mold fungi, and actinomycetes). All methods are presented in Table S3. The in-house microbial analysis was performed on all samples, while the more detailed microbial analysis conducted by the service laboratory encompassed the sand and GAC influents and effluents from S:t Olof WWTP, and the GAC influent and effluent from Kivik WWTP. Samples for the accredited service laboratory were stored in 500 mL bottles that contained 10 mg thiosulfate. All samples were stored in the dark at 6–9°C and sent within 24 h, except those from the first sampling occasion, which were sent within 4 days.

E. coli and total coliform bacteria were analyzed within 48 h from sampling, with IDEXX Colilert 18 and Quanti-Tray 2000, per the manufacturer's instructions. The influents from both WWTPs were diluted 1:100,000 with drinking water prior to analysis. The sand and GAC influents from S:t Olof WWTP were diluted 1:1,000.

E. coli and total coliform concentrations were measured both internally, on the day of the experiments, and by the external laboratory, allowing for control of potentially deviating results due to, for example, continued growth of the bacteria. No substantial deviations were observed (Table S2).

TCC was analyzed using flow cytometry (FCM) on a BD Accuri C6 Plus. The influents to both WWTPs were passed through a 10 μm filter prior to analysis to protect the instrument from clogging and were then diluted 1:10 due to high cell counts. SYBR^® Green I (dilution: 100× in dimethyl sulfoxide and final concentration in samples: 1×) was used to measure TCC, and the samples were incubated at 37°C for 15 min after staining. Fifty microliter of sample was analyzed at a flow of 35 mL/min (except for the influent to S:t Olof WWTP, for which 25 μL of sample was analyzed on the last occasion due to high cell counts). The samples were analyzed in duplicate and the average of the resulting values was used for the data analysis. The flow cytometric analysis was conducted within 24 h from sampling.

The chemical analysis was performed in-house (dissolved organic carbon (DOC), ⁠, ⁠, ⁠, ⁠, and conductivity). Additional chemical analysis of 114 parameters in the Swedish drinking water legislation was performed on one sample (Kivik 400 J/m² from October 2). This sample was selected based on previous experiments and microbial analysis and the additional chemical analysis was conducted at the same accredited laboratory as the microbial analysis (Eurofins; the methods are presented in Table S3). In total, 400 J/m² is a commonly used UV fluence for drinking water production in Sweden (Saguti et al. 2022) and the sample thus represents a standard scenario.

In-house analysis of DOC, nitrogen, and phosphorus fractions were performed within 24 h from sampling. The samples were filtered (0.45 μm, GVS filter technology) and stored in the dark at 6–9°C prior to analysis. The nitrogen and phosphorus fractions were analyzed on a Metrohm Eco Ion chromatograph and DOC was analyzed using a Shimadzu (TOC-L). In addition, turbidity and suspended solids were measured, resulting in 132 parameters that were analyzed in the Kivik 400 J/m² sample.

GAC filtration increases total bacteria concentrations (Wilcox et al. 1983; Miller et al. 2020; Takman et al. 2023) but has been shown to selectively decrease the concentrations of E. coli and total coliform bacteria (El-Zanfaly et al. 1998; Hijnen et al. 2010; Spit et al. 2022), as observed in our study. Examining only the influents to and effluents from the GAC filters, the LRV was 0.7 for E. coli and 1.0 for total coliform bacteria at Kivik WWTP, versus 0.4 for both E. coli and total coliform bacteria at S:t Olof WWTP – demonstrating a similar but slightly higher removal in the GAC filter at Kivik WWTP. Many factors could contribute to this pattern, such as differences in operational conditions (for example, EBCT, Table S1) or in upstream treatment processes, causing variations in nutrient, carbon, and bacterial loads and composition.

The E. coli and total coliform concentrations in both WWTP effluents exceeded the limits defined in the Swedish drinking water criteria (present in 100 mL) (LIVSFS 2022:12). The average E. coli concentration in the Kivik WWTP effluent was 2 cfu/100 mL, meeting the quality class A criteria (<10 cfu/100 mL) for irrigation per Regulation (EU) 2020/741 of the European Parliament and of the Council of 25 May 2020 on minimum requirements for water reuse. The average effluent E. coli concentration from S:t Olof WWTP was 1967 cfu/100 mL, exceeding the limits for quality class C water on all occasions and meeting the criteria for quality class D.

Quality class A water can be used for the irrigation of ‘all food crops consumed raw where the edible part is in direct contact with reclaimed water and root crops consumed raw’ whereas water that is deemed quality class D can be used for irrigation of ‘industrial, energy and seeded crops’ (Regulation (EU) 2020/741, Annex I).

Previous studies (Takman et al. 2023) have reported that the concentrations of E. coli in the effluent from the MBR + GAC process at Kivik WWTP sometimes failed to pass the class A criteria, meeting only quality class B standards. However, these measurements were conducted in 2021 and more recent measurements conducted by the water utility during 2023 and 2024 show that concentrations for quality class A were achieved.

The average effluent turbidity was ≤1 NTU and the suspended solids concentration in the effluent was <1 mg/L for both WWTPs (Table 2). The effluent conductivity was 787 μS/cm at Kivik WWTP and 649 μS/cm at S:t Olof WWTP.

Table 2

General performance parameters: Kivik and S:t Olof WWTPs

.	DOC .	.	.	.	.	Suspended solids .	Turbidity .	Conductivity .
	mg/L .	mg N/L .			mg P/L .	mg/L .	NTU .	μS/cm .
Kivik in MBR	22.9	29.7	2.8	0.3	3.0	52	25
Kivik in GAC	4.6	0.04	23.6	nd	0.2	<1	<1
Kivik effl GAC	3.5	0.04	25.5	0.06	0.2	<1	<1	787a
S:t Olof in CAS	37.5	24.5	3.8	0.6	2.9	191	161
S:t Olof in sand	6.5	1.2	20.5	0.45	ndb	7	3
S:t Olof in GAC	5.9	0.8	21.6	0.03	ndb	1	1
S:t Olof effl GAC	4.5	0.6	22.2	0.05	ndb	<1	<1	649a

.	DOC .	.	.	.	.	Suspended solids .	Turbidity .	Conductivity .
	mg/L .	mg N/L .			mg P/L .	mg/L .	NTU .	μS/cm .
Kivik in MBR	22.9	29.7	2.8	0.3	3.0	52	25
Kivik in GAC	4.6	0.04	23.6	nd	0.2	<1	<1
Kivik effl GAC	3.5	0.04	25.5	0.06	0.2	<1	<1	787a
S:t Olof in CAS	37.5	24.5	3.8	0.6	2.9	191	161
S:t Olof in sand	6.5	1.2	20.5	0.45	ndb	7	3
S:t Olof in GAC	5.9	0.8	21.6	0.03	ndb	1	1
S:t Olof effl GAC	4.5	0.6	22.2	0.05	ndb	<1	<1	649a

Note: Internal analysis. Average of all three sampling occasions.

^aMeasured on 2023-10-02.

^bnd = not detected.

View Large

The differences in bacterial removal in the GAC and sand filters could be due to differences in the composition of the biofilm, which seem to prevent the growth of C. perfringens and intestinal enterococci in the Kivik GAC filters to a greater extent than in the S:t Olof sand and GAC filters. The biofilm composition and function may be affected by several factors, such as the influent concentration and composition of microorganisms, DOC, and nutrients – and thus by the upstream treatment processes (Li et al. 2014; Vignola et al. 2018; Dottorini et al. 2021).

The concentrations of culturable microorganisms (that is, not the total number of cells as determined by FCM) were higher in the sand and GAC influents at S:t Olof WWTP compared with the GAC influent at Kivik WWTP, likely due to the UF process upstream of the GAC filters at Kivik WWTP (Table 3). All microbial concentrations (except that of Actinomycetes, which were consistently ≤1 cfu/100 mL) decreased over the Kivik GAC filter, whereas some concentrations increased over the S:t Olof GAC filter (Table 3). The concentrations in the GAC effluents of most microbial parameters exceeded the limits in the Swedish drinking water criteria (LIVSFS 2022:12) (Table 3). The GAC filter downstream of the MBR removed E. coli, total coliform bacteria, C. perfringens, and intestinal enterococci to a higher degree than the GAC filter downstream of the CAS + sand process. This indicates that the processes upstream of GAC filters influence their removal capacity and that GAC filters perform better, regarding removal of microbial contaminants, downstream of an MBR versus a CAS + sand process. The effluent microbial concentrations, however, demonstrate that disinfection is necessary to ensure a high microbial water quality.

In the GAC effluents, the concentration of Actinomycetes was the only microbial parameter in the Swedish drinking water criteria (LIVSFS 2022:12) that was met on all occasions (Table 3). The limits for C. perfringens, E. coli, total coliform bacteria, intestinal enterococci, and microfungi (sum of yeast and mold fungi) were exceeded in several samples. Thus, these parameters were analyzed in the UV disinfection experiments. As the microbial water quality monitoring conducted in this study was only based on a few occasions, it is not possible to draw any conclusions on ‘unusual changes’ for the HPC and slow-growing bacteria.

A high inactivation, that is, a disabling of growth, of E. coli, total coliform bacteria, and intestinal enterococci in water can be achieved with UV radiation (Sommer et al. 2000; Jacangelo et al. 2003; Mezzanotte et al. 2007). Similar results were yielded in the present study, in which a UV fluence of 200 J/m² was sufficient to obtain concentrations <1 cfu/100 mL for these microbial indicators in the Kivik (MBR + GAC) and S:t Olof (CAS + sand + GAC) effluents, meeting quality class A criteria for irrigation and Swedish drinking water criteria (Table 4). Inactivation of C. perfringens and microfungi requires higher UV fluence (Hijnen et al. 2006; Carabias et al. 2023), consequently, in our study, the concentration of C. perfringens was not consistently lowered to <1 cfu/100 mL by any UV fluence in the CAS + sand + GAC effluent. The concentrations in the MBR + GAC effluent before UV exposure, however, were low (≤2 cfu/100 mL) and a UV fluence of 200 J/m² was sufficient to achieve concentrations that were consistently <1 cfu/100 mL.

Microfungi were detected in the effluents from both WWTPs after exposure to all UV fluences, but following exposure to UV fluences >400 J/m², these concentrations did meet the drinking water criteria (<100 cfu/100 mL). The HPC and slow-growing bacteria concentrations in the CAS + sand + UV effluent (S:t Olof WWTP) varied between <1 cfu/mL and approximately 160 cfu/mL, corresponding to a range of at least two orders of magnitude. This variation could be considered wide, but since the concentrations of C. perfringens and microfungi in these samples exceeded the drinking water limits, the exact definition of ‘no unusual change’ is not crucial for the conclusions.

In the Kivik samples exposed to 200 J/m² fluence, the limit for microfungi was exceeded on the last occasion. The concentration of 550 cfu/100 mL is, however, difficult to explain with the concentration from the upstream GAC effluent of 37 cfu/100 mL. It seems likely that the sample was contaminated during transport or analysis, or that the fungi were able to continue growing between the time of the sampling and the analysis. Due to technical errors by the external laboratory, there was no analysis reported for slow-growing bacteria in the Kivik 200 J/m² sample from June 12, 2023. Other than the high microfungi concentration in the Kivik 200 J/m² sample, all concentrations in the UV-treated MBR + GAC effluent samples (Kivik WWTP) met the Swedish drinking water criteria after exposure to all UV fluences. The TCC in the Kivik GAC effluent varied between 500,000 and 900,000 cells/mL. While in the higher range of TCC reported for drinking water, this is not unprecedented (Schleich et al. 2019; Rosenqvist et al. 2023). All cells in the water are affected by, and absorb, UV irradiation, and the results therefore confirm that the studied UV fluences are sufficient to treat the WWTP effluents to the extent that they meet microbial drinking water criteria also with these cell concentrations. It should be noted, however, that the hydraulics in the UV reactor, including variables such as the orientation of the reactor, mixing, and potential turbulence, can affect the disinfection efficiency (Ho et al. 1998). Thus, additional assessments of the impact of different UV fluences at full scale will be required to validate the results reported here for laboratory-scale UV disinfection.

Treatment of the effluent from Kivik WWTP (MBR + GAC) with UV fluence >400 J/m² resulted in water that fulfilled the Swedish criteria for microbial drinking water quality and thus, one of these samples (from October 2, 2023) was selected for extensive chemical analysis, comprising 114 parameters from the Swedish criteria (the results are presented in Table S4). The parameters included metals, pesticides, per- and polyfluoroalkyl substances (PFAS), trihalomethanes (THMs), and polycyclic aromatic hydrocarbons.

The concentrations of all chemical parameters in the UV-treated effluent met the drinking water criteria, except nitrate (Table S4). In the effluent from an activated sludge process with denitrification, however, the nitrate concentration would likely meet the criteria. A selection of the chemical parameters is presented in Table 5, together with their limits and indicator values.

The concentration of PFAS 4 (2.9 ng/L, the sum of perfluorooctane sulfonate (PFOS), perfluorooctanoate (PFOA), perfluorononanoate (PFNA), and perfluorohexane sulfonate (PFHxS); Table S4) approached the limit (4.0 ng/L) in the Swedish criteria. Given the substantial seasonal variations in flow previously reported at Kivik WWTP (Takman et al. 2023) the concentrations of all water quality parameters will likely vary throughout the year, potentially causing concentrations that exceed the criteria. The examined sample was collected during a period of low flow (Figure S1), and it is thus likely that contaminants in the water will generally be more diluted during the remainder of the year, which would decrease their concentrations. Nevertheless, several samples, collected over a longer period, should be examined with a similar chemical panel to ensure that concentrations of key parameters are consistently meeting the drinking water criteria.

In addition, there are no limits on the levels of pharmaceuticals in Swedish drinking water criteria. These substances could be present in higher concentrations in wastewater compared with groundwater and surface water, which are currently used in conventional drinking water production in Sweden. Concerning the potable reuse of wastewater, it should, therefore, be examined whether the drinking water criteria need to be updated with such compounds. Drinking water quality is further defined as follows (translated from Swedish):

Full-scale GAC filtration at two WWTPs, followed by UV disinfection conducted at the laboratory scale, was examined to determine the influence of upstream treatment (MBR and CAS followed by sand filtration) on the GAC and UV treatment capacities. The implications of this on wastewater reuse were analyzed by comparing effluent water quality to criteria for wastewater to be reused for irrigation, and to drinking water criteria. Three UV fluences (approximately 200, 400, and 700 J/m²) were evaluated, and based on the UV laboratory experiments, one sample (MBR + GAC effluent treated with a UV fluence of 400 J/m²) was selected for extensive chemical analysis, comprising over 100 parameters in Swedish drinking water legislation. The following conclusions were drawn:

• E. coli, total coliform bacteria, C. perfringens, and intestinal enterococci were removed to a higher extent in the GAC filter that followed downstream of an MBR, compared to the GAC filter that followed downstream of a CAS process and sand filtration.

• Based on an extensive microbial and chemical analysis (approximately 130 parameters), drinking water quality was achieved per Swedish criteria in the MBR + GAC effluent that was treated with a UV fluence of 400 J/m², with the sole exception being nitrate. High nitrate concentrations can be mitigated with a biological wastewater treatment process with denitrification.

• With the two higher UV fluences (400 and 700 J/m²), microbial drinking water criteria were met after UV disinfection of the MBR + GAC effluent.

• Without UV disinfection, the MBR + GAC effluent met the criteria for quality class A water for irrigation (Regulation (EU) 2020/741) on all occasions, whereas the CAS + sand + GAC effluent met the criteria for quality class D water.

• The concentrations of E. coli, total coliform bacteria, and intestinal enterococci were <1 cfu/100 mL at UV fluences ≥200 J/m², regardless of the upstream treatment processes.

The project was conducted within the ÅÄÖ Project, which was funded by Formas – a Swedish Research Council for Sustainable Development (No. 2021-02656). Further, the Agenda 2030 Graduate School at Lund University is gratefully acknowledged for its financial support. The authors thank Ronny Helgesson at Österlen VA for assistance with sampling.

The project was funded by Formas – a Swedish Research Council for Sustainable Development (No. 2021-02656), and by the Agenda 2030 Graduate School at Lund University. Formas and the Agenda 2030 Graduate School had no involvement in the study design, the collection, analysis or interpretation of data, the writing, or in the decision to submit the article for publication.

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

The authors declare there is no conflict.