Methodology for an ecological solution of subsurface flow constructed wetlands used in the treatment of greywater

Hammams or public baths are essential sanitary establishments in the social life of North African and Middle Eastern countries. In Morocco, hammams are part of the cultural heritage, with a number of units of about 12,000 distributed throughout the country (AMEE 2019). They consume enormous quantities of water and obviously discharge equivalent quantities of wastewater, such that the average consumption varies from 60 to 120 m³/day. Moreover, the water used in these hammams can be drinking water or well water (Hajji et al. 2008; Chakri et al. 2019). On the other hand, about 844 million people still need a basic supply of drinking water and about 159 million others use it directly from surface water resources (Bajpai et al. 2019). This dangerous situation, therefore, requires a practical, economical and ecological solution for the reuse of wastewater discharged from hammams, especially since this water is grey and rich in biodegradable organic matter (Chakri et al. 2019), which reflects a low degree of pollution compared to black water.

Different technologies exist in the field of decentralized greywater treatment. Membrane systems such as the membrane bioreactor (MBR) represent a small footprint solution that is characterized by a high treatment load and very good effluent quality, but it is limited by high investment, maintenance and energy consumption costs. Another technology close to the previous one is the biomass concentrator reactor (BCR), which uses high porosity filters, consumes less energy and produces better-treated effluents than the MBR. However, this technology is in development for commercial use (Capodaglio et al. 2017; Bajpai et al. 2019). An improvement in membrane technology is the use of hybrid systems that combine MBR and BCR to control filter fouling by introducing electrical fields (Bajpai et al. 2019). Small-scale wastewater treatment plants (WWTPs) are also adopted in greywater treatment using anoxic or aerated tanks with some sedimentation, which makes them quite sensitive to transient hydraulic loads and changes in sludge quality. Although they give a moderate effluent quality, anaerobic treatment systems (UASBs) provide a biological treatment of greywater that allows the production of methane and biomass, as they are usually followed by another treatment system in order to obtain the desired effluent quality (Capodaglio et al. 2017).

In general, hammam owners cannot implement decentralized treatment systems such as the MBR because they are too expensive and consume a lot of electricity. Compared to conventional wastewater treatment technologies, constructed wetlands (CWs) are inexpensive, robust, simple to use and can be constructed from locally available materials (Wallace & Knight 2006). Therefore, an efficient decentralized treatment can be ensured using CWs to obtain effluents that can be used in the irrigation of certain types of crops, toilet flushing, fire protection and other applications (Bajpai et al. 2019).

Subsurface flow CWs (SSF CWs) are effective for total suspended solids (TSS) removal and biological oxygen demand (BOD) due to the relatively low flow velocities and high fluid surface area (US EPA 2000), which coincides with the nature of greywater discharged from hammams. Vertical flow CWs (VF CWs) are much more aerobic than horizontal flow CWs (HF CWs) and offer suitable conditions for nitrification. On the other hand, VF CWs do not provide any denitrification (Vymazal 2005). Compared to HF CWs, VF CWs require less land, typically 1–3 m²/PE, while HF CWs require 5 m²/PE (Vymazal 2010). In general, VF CWs are expected to perform better than HF CWs for BOD reduction because VF CWs are intermittently loaded and have an unsaturated flow resulting in higher oxygen transfer to the filter media compared to HF CWs (Zhang et al. 2014).

Several models exist in the literature for the dimensioning of CWs, which are often considered black boxes and therefore difficult to model their behaviors (Rousseau et al. 2004). The most common design approaches are ground rules, regression equations, influent loading graphs, k-C* model, P-k-C* model and oxygen demand (Platzer 1999; Rousseau et al. 2004; Wallace & Knight 2006; Dotro et al. 2017). Sometimes a combination of two or three approaches can be adopted, as in this work, where the influent loading graphs are combined with the P-k-C* model to calculate the required surface of the HF CW. Thus, they were also combined with the oxygen demand approach to determine the required area of the VF CW. However, Kadlec (2000) has shown that the k-C* model is inadequate for the design of CWs, resulting in reduced use by engineers. The basic rules and regression equations are not used here, as they were developed under specific conditions such as site climate, pretreatment technology, wastewater type and other factors (Dotro et al. 2017).

There is a lack of practical studies that provide engineers and practitioners with a methodology for designing SSF CWs. This implies a disruption in the design methods, as each one has been developed based on local parameters such as the nature of the greywater source and the climate where the system will be implemented. This will lead to a big difference in the results obtained from local conditions compared to others. The lack of this type of study ultimately results in poorly designed systems that do not meet expectations and have poor treatment performance. Hence the novelty of the present study, which sets up a practical and flexible methodology that can be standardized in any climate and location, regardless of the nature of the greywater source.

This work proposes a practical method for sizing VF and HF CWs using a hybrid system for the treatment of greywater discharged from hammams (public baths), including suggestions and feedback from experts in the field. The quantification of water evapotranspiration (ET) represents an essential step in the design of CWs in SSF, but it has become ignored by most designers due to the complexity of establishing it theoretically and especially in the cases of new installations where it is impossible to measure it experimentally according to the real data that characterize the implemented system. This has a negative impact on the calculation of the necessary surfaces of the SSF CWs, which comes mainly from the water balance that does not take into account the ET. For these reasons, the quantification of ET will not be ignored here, as a semi-empirical method will be presented to quantify the water ET through SSF CWs. A calibration will thus be ensured at the water balance level which leads to a reliable and close-to-reality design. The main objective of this paper is:

• To provide a practical and flexible methodology for designers of SSF CWs intended for the treatment of greywater, especially those discharged by hammams.

• To solve the ET quantification problem for SSF CW design.

• To provide designers with a concrete case study using this methodology.

To generate the present methodology, a revision of previous works was elaborated taking into account those characterized by the practical theme of designing SSF CWs intended for greywater treatment (Brix & Arias 2005; Kadlec & Wallace 2009; Dotro et al. 2017). Deficiencies have been detected in the integrity and flexibility of the proposed methodologies, such as they are sometimes limited only to the phase of establishing the required surfaces of the CWs without taking into account the ET, while others are based on specific conditions of the climate and the source system of the generated methodology that can never be used in other conditions. A new flexible and integrated methodology has been developed by combining and improving the previous ones by collecting relevant recommendations from experts in design practice. The different steps of the proposed methodology will be presented in the following article.

The first step in designing CWs is to characterize the site where the treatment system will be installed. For this purpose, it is essential to define the area reserved for the installation in order to compare it with the calculated area. Indeed, the feasibility of the treatment system can be well discussed. The location of the site is also necessary to know the type of climate and the meteorological data. The water losses and gains at the CW level can be quantified through the water balance, as well as the choice of vegetation adapted to the climate of the site.

The characterization of the influents discharged by hammams represents the basis of all that follows. The objective is to know the nature of these waters as well as the daily water flow. The type of SSF CW (VF or HF or both) most suitable for the measurement results of the physico-chemical and biological characteristics of the wastewater can be chosen.

Several studies have focused on the characterization of greywater, including water discharged from private and public baths (Eriksson et al. 2002). In Morocco, a study was carried out on greywater discharged by 15 Hammams in the Dar Bouazza region. Table 1 presents the results of the measurements obtained and the limit values imposed by the water quality standards for some countries.

Table 1

Characteristics of the influents of 15 hammams in the region of Dar Bouazza in Morocco with the standards intended for irrigation

.	.			Water standards for irrigation .
.	Measurement results (Chakri et al. 2019) .			S.E.E.E (2007)  .	Jeong et al. (2016) .
Parameters .	Min .	Max .	Average .	Morroco .	US EPA .	Cyprus .	Italy .	France .	Spain .	Greece .
T (°C)	27.3	35.5	30	35	–	–	–	–	–	–
Conductivity (⁠S/cm)	–	–	>2,669	12,000	–	–	3,000	–	–	–
BOD5 (mg/L)	10	410	167	–	30a	30b	20	–	–	25c
COD (mg/L)	64	836	385.8	–	–		100	60d	−	–
TSS (mg/L)	35	724	431	100	30a	45	10	15d	35b	35c
TKN (mg/L)	2.8	86.9	16.22	–	–	–	–	–	–	–
Phosphate (mg/L)	–	–	–	–	–	–	5	–	–	–
Sulfate (mg/L)	–	–	–	250	–	–	–	–	–	–

.	.			Water standards for irrigation .
.	Measurement results (Chakri et al. 2019) .			S.E.E.E (2007)  .	Jeong et al. (2016) .
Parameters .	Min .	Max .	Average .	Morroco .	US EPA .	Cyprus .	Italy .	France .	Spain .	Greece .
T (°C)	27.3	35.5	30	35	–	–	–	–	–	–
Conductivity (⁠S/cm)	–	–	>2,669	12,000	–	–	3,000	–	–	–
BOD5 (mg/L)	10	410	167	–	30a	30b	20	–	–	25c
COD (mg/L)	64	836	385.8	–	–		100	60d	−	–
TSS (mg/L)	35	724	431	100	30a	45	10	15d	35b	35c
TKN (mg/L)	2.8	86.9	16.22	–	–	–	–	–	–	–
Phosphate (mg/L)	–	–	–	–	–	–	5	–	–	–
Sulfate (mg/L)	–	–	–	250	–	–	–	–	–	–

^aProcessed food crops.

^bCrops for human consumption.

^cRestricted.

^dUnrestricted.

View Large

In general, the limit value of water intended for irrigation is 30 mg/L concerning the parameters BOD₅, COD and TSS. Most of the water discharged by hammams is grey with traces of black water. The ratios COD/BOD₅ and BOD₅/COD reflect the richness in organic matter as well as the biodegradable nature of hammam wastewater (Chakri et al. 2019), but unfortunately, Chakri et al. (2019) did not indicate the concentrations of phosphate and sulfate even though they are important parameters in the analysis of greywater quality.

The choice of treatment system configuration using CWs depends primarily on the nature of the water being treated, geographic location, cost, land availability and treatment objective (Horner et al. 2011). For hammam discharge, a hybrid system combining VF and HF CWs is the best solution for organic matter and nitrogen removal, which allows a higher treatment effect to be achieved using the advantages of each system individually (Vymazal 2005). Generally, VF CWs are followed by HF CWs (Vymazal 2005; Kadlec & Wallace 2009), the configuration adopted here. Because of the constraint of needing rest periods for the VF CWs, three VF CWs in parallel ensure a continuous operation of the overall system, such that each bed operates for 2 days and rests for 4 days, as well as good system flexibility. Several experiments have shown that the performance of VF CWs is strongly dependent on the pretreatment (Kadlec & Wallace 2009). Before feeding the VF CW, an efficient primary treatment is, therefore, necessary to remove solid particles in order to avoid filter clogging (Dotro et al. 2017). In addition, the range of characteristics presented in Table 1 confirms the use of pretreatment. For SSF CWs, the use of lagoons is not the preferred method for primary treatment (US EPA 2000). The pretreatment achieved by the septic tank is important to ensure the success of the secondary treatment achieved by the CWs and to avoid clogging of the filters (Brix & Arias 2005). More than 45% of the final treatment can be done in the septic tank (Bounds 1997) (Figure 1).

Generally, the septic tank is sized with two compartments (UN-HABITAT 2008). The first of which is larger and allows enough time for the sludge to settle to the bottom and a scum layer to float to the surface, while the second is designed to settle a small amount of solids. A baffle is used to separate the two compartments with an opening in the center through which water passes from the first compartment to the second (Adhikari & Lohani 2019). The following are typical recommendations for septic tank design (Bounds 1997):

• The path of travel between the entrance and exit must be longer than the width or depth.

• The ratio of length-to-depth (L:D) of liquid varies from 1:1 to 3:1 (1.5:1 to 2.5:1 is the most common).

• The typical height-to-width (H:W) ratio is 1:1.

The ET increases the concentration of pollutants due to the loss of pure water. It temporarily lowers the water level, increases the hydraulic retention time (HRT), and decreases the hydraulic loading rate (HLR) (US EPA 2000; Wallace & Knight 2006). For this reason, it is important to quantify the ET but it is difficult to calculate it accurately in small SSF CWs (Dotro et al. 2017), which is the case in this study. L’ET is highly dependent on plant type and density, climatic conditions of the site, height/width ratio (US EPA 2000), orientation relative to the general wind direction and the growing season of the plants (Dotro et al. 2017).

is the dimensionless crop coefficient and is the reference evapotranspiration (mm/day). The FAO-56 Penman–Monteith method is commonly used to calculate the ⁠. The free software ET0, provided by FAO-56, is used to calculate the ⁠, which corresponds to the CW site. Several analytical methods exist in the literature to estimate and each has advantages and disadvantages. The method used in this study takes into consideration the type of crop, the variation of crop growth periods, the length of each period, the crop height and the climatic conditions of the site. According to FAO-56, each plant has its own crop coefficients appropriate for each growing period (Bos et al. 2009). Table 2 lists some of the plants used in CW.

Table 2

Seasonal crop coefficients by plant type (Bos et al. 2009)

Plants .	Initial period .		Mid-season period .		End of season period .
	.	(d) .	.	(d) .	.	(d) .
Cattails, Bulrushes, killing frost	0.30	180	1.20	90	0.30	35
Cattails, Bulrushes, no frost	0.60	–	1.20	–	0.60	–
Short veg., no frost	1.05	180	1.20	90	1.10	35
Reed swamp, standing water	1.00	–	1.20	–	1.00	–
Reed swamp, moist soil	0.90	–	1.20	–	0.70	–

Plants .	Initial period .		Mid-season period .		End of season period .
	.	(d) .	.	(d) .	.	(d) .
Cattails, Bulrushes, killing frost	0.30	180	1.20	90	0.30	35
Cattails, Bulrushes, no frost	0.60	–	1.20	–	0.60	–
Short veg., no frost	1.05	180	1.20	90	1.10	35
Reed swamp, standing water	1.00	–	1.20	–	1.00	–
Reed swamp, moist soil	0.90	–	1.20	–	0.70	–

View Large

is the crop coefficient provided by FAO-56, is the wind speed (m/s), is the minimum relative humidity (%) and is the height of the plant (m). Generally the adjustments of are made using the average values for and during the mid-season period (and also at the end of the season if K_{c end} > 0.45) as well as a maximum value of the plant height to have conservative results. K_{c ini} is corrected using graphs that take into account ET₀, soil type and wetting frequency (Bos et al. 2009) (Appendix 1, Supplementary material).

Generally, there is no problem in estimating precipitation (PR) because it is derived from weather data provided by the weather stations closest to the study site.

is the wastewater input rate (L/day), is the COD removal efficiency (%) (Platzer (1999) measured 85%), is the chemical oxygen demand of the influents (g/L), is the influent total Kjeldahl nitrogen concentration (g/L) and is the influent nitrate concentration (g/L).

The next step is to use the influent loading rate figures to verify the effluent outlet concentration using the previously calculated area. We recommend using relative conservatism of 50% as a basis for evaluation (Appendix 2, Supplementary material). If the degree of conservatism is low (less than 50%), the OTR should be adjusted by decreasing it to increase the required surface area, and therefore the effluent outlet concentration will decrease while the degree of conservatism will increase. On the other hand, if the degree of conservatism is very high (greater than 90%), it must be decreased by increasing the OTR to decrease the required surface area. It should be noted that VF CWs generate lower effluent concentrations than those shown in the influent loading rate figures (Wallace & Knight 2006).

VF CWs are typically constructed with a pre-specified bed depth (h). Several recommendations exist in the literature; such as 50 < h < 80 cm (Kadlec & Wallace 2009) and 30 < h < 70 cm (Stefanakis et al. 2014). The maximum value of the bed depth can be up to 100 cm (Brix & Arias 2005).

In practice, most treatment CWs are designed to remove multiple pollutants (Dotro et al. 2017). For this reason, we recommend performing the calculations for all pollutants of interest, the designer then selects the resulting surface that will remove all pollutant targets.

After calculating the surface area, it must be ensured that the effluent concentration is below the limit value using the influent loading rate graphs. If the degree of conservatism is low (less than 50%), the reaction coefficient should be adjusted should be adjusted by increasing its conservatism to increase the area required to achieve the required output concentrations at a conservatism level of 50% or greater.

Several recommendations exist in the literature for the depth h of HF CWs, such as 27 < h< 60 cm (Wallace & Knight 2006; UN-HABITAT 2008), 50 < h < 70 cm (Dotro et al. 2017) and 60 < h < 80 cm (Vymazal 2018).

For a VF CW, the aspect of L/W ratio is not as important as for a HF CW, considering that the power supply is done from the top of the filter. However, for a HF CW, the designer is confronted with two contradictory design objectives (Kadlec & Wallace 2009):

• The desire to spread the influent over a very large flat area, this delays the clogging time.

• The desire to spread the influent as evenly as possible over the vertical cross-section improves flow distribution.

As the L/W ratio increases from 1/1 to 4/1, the removal efficiency of COD and ammonia nitrogen (⁠⁠) becomes higher, while that of sulfate sulfur (⁠⁠) becomes lower (Sanchez-Ramos et al. 2017). Several recommendations of typical ratios are proposed in the literature. In practice, it is advisable to avoid a ratio that exceeds 4/1 because it can cause hydraulic problems (US EPA 2000).

is the wastewater flow rate in the HF CW (L/day), is the input concentration of BOD (g/L), and is the cross-sectional area of the HF CW (⁠ (m²). Vymazal (2018) recommends using the total area of the HF CW (⁠⁠, such that OLR is less than 10 g BOD₅/m²/day, 10 g TSS/m²/day and 20 g COD/m²/day.

is the wastewater flow entering the VF CW (L/day), is the chemical oxygen demand input concentration (g/L) and is the cross-sectional area of the VF CW (⁠ (m²). If these conditions are not validated, the bed cross-sectional area must be adjusted. For HF CWs, the bed width and/or depth is increased, and for VF CWs the bed width and/or length is increased. Also proposed is another method, which consists in evaluating the Damköhler number (Da), such that Da < 0.09 indicates that there is a high probability of clogging the wetland. In contrast, Da > 0.09 represents less of a tendency for the filter to become clogged. This method can be applied in HF and VF CWs (Austin et al. 2007; Stefanakis et al. 2014), but it requires experimental data to apply it.

In HF CWs, the goal is to achieve a uniform distribution over the entire cross-section of the inlet end of the bed. The most common method of water distribution is the use of perforated plastic pipes, usually reinforced polyethylene with a buried location in the inlet area and the surface of the inlet area that is leveled with a filter bed. The holes should be at least 3 cm in diameter to avoid uneven water distribution due to clogging of the holes, and they should be regularly spaced at a distance approximately equal to 10% of the width of the cell. Effluent collection in HF CWs can be achieved by a simple perforated pipe buried at the bottom of the outlet area. This leads to a sump located just after the HF CW to control the water level in the HF CW either by a socket pipe, a swivel elbow, or flexible pipes that can be held in position by a chain or a rope (Vymazal & Kröpfelová 2008; Kadlec & Wallace 2009).

Most VF CWs are pulse fed. This method involves storing influent water and then rapidly dosing it onto the bed surface, either through the use of pumps, or through gravity drains (if there is an adequate drop on site) or siphons. The design goal is to rapidly flood the bed in less than 15 min with shorter dose intervals; the 5–10 min range is more desirable. There is no rational design method for volume per dose (Kadlec & Wallace 2009). For VF CWs, one of the most common distribution methods is a series of perforated influent distribution pipes placed above the surface of the CW bed, with orifices at least 8 mm in diameter and spaced 0.4–0.7 m (Brix & Arias 2005; Dotro et al. 2017). Water collection is done through a network of perforated drainage pipes placed at the bottom of the bed. This network is connected to the atmosphere via vertical aeration pipes, which extend above the bed and allow a better aeration of the deepest parts of the bed, as well as the distance between the pipes, which is usually less than 5 m but it depends on the size of the bed (Stefanakis et al. 2014). The collection system is usually oversized to allow passive aeration of the VF CW bed. They are most commonly designed as free-draining systems, and water level control is possible and may be desirable, but is not mandatory for these systems (Kadlec & Wallace 2009).

When selecting the vegetation type for any CW type, designers should consider the following criteria:

• The plants chosen must be available on the local market, and adapted to the climatic conditions of the site.

• They must be tolerant to a variety of pollutants present in the wastewater (organic matter, nitrogen, phosphorus…) with a high removal capacity (direct absorption or indirect by providing adequate conditions such as oxygen transfer).

In SSF CWs (VF or HF), emergent macrophytes are generally used extensively, such as Phragmites, Typha, Scirpus and Schoenoplectus, which represent the most frequently used plant species (Stefanakis et al. 2014).

Finer materials may result in better treatment, but may cause hydraulic problems and clogging. On the other hand, coarser materials allow water to flow easily through the wetland bed (Wallace & Knight 2006). The type of wastewater can also be an important factor in the selection of media size, which increases from greywater through secondary to primary influents (Kadlec & Wallace 2009). For HF CWs, using coarse material in the first part of the HF CW bed and finer material closer to the outlet provides the best treatment with ease of flow. The minimum media size is 4 mm, while the media in the outlet and inlet areas should be between 25 and 50 mm in size (Wallace & Knight 2006). Vymazal (2018) recommends higher sizes ; between 5 and 20 mm in the main layer, while the size of the media in the inlet and outlet areas is between 50 and 200 mm. Usually, different layers of gravel with an increasing grain size from top to bottom are used for VF CWs. In general, the bed is divided into three layers; a sand configuration, an intermediate layer of medium-sized gravel, and a drainage layer of larger gravel (Stefanakis et al. 2014).

The bed slope allows for the collection of treated water and drainage to the outlet of the area. However, there is a paucity of research to determine the optimal slope (UN-HABITAT 2008). In general, sloped HF CWs have shown relatively high treatment efficiency compared to non-sloped HF CWs (Hua et al. 2021). A slope of 0.5–1% is recommended for ease of construction and proper drainage (US EPA 2000).

The treatment system configuration will be as recommended in this methodology as it fits within the framework of treating greywater discharged from public baths.

Table 4 shows the calculated septic tank dimensions using 24 h as the HRT to ensure effective TSS removal.

Table 4

Septic tank calculation results

Input parameters .	Output parameters .
Flow of wastewater to be treated:	Total volume: ;
Hydraulic retention time:	Volume of the room 1: ⁠; 1.33
	Volume of the room 2: ; 0.66

Input parameters .	Output parameters .
Flow of wastewater to be treated:	Total volume: ;
Hydraulic retention time:	Volume of the room 1: ⁠; 1.33
	Volume of the room 2: ; 0.66

View Large

The surface area of the VF CW was calculated by applying the Platzer model, such that a conservative percentage of attenuation in the septic tank was taken into account; 30% of organic matter (COD and BOD₅) and 50% of TSS, however the attenuation of TKN and NO_3-N which represents denitrification was eliminated to have a conservative calculated surface area. The input parameters and calculation results are summarized in Table 5.

The influent loading graph confirm the validity of the calculated VF CW area (40.5 m²). This provides sufficient enough output concentrations of BOD₅ and TSS as an intermediate treatment step, such that the removal efficiencies are 66.15 and 73.07%, respectively, while the resulting output concentration at the TKN loading rate is higher than the input (41.38 > 16.22 mg TKN/L). The OLR_COD is higher than 27 g COD/m²/day, this may cause a risk of bed clogging. However, the presence of three VF CW in parallel avoids this problem, as two beds can be fed at the same time, and the OLR_COD will be reduced to half its value.

The output concentrations of BOD₅ and TSS determined from the influent loading graphs (26.86 and 29. 17 mg/L) are lower than the limit value of irrigation water (30 mg/L), as well as the calculated OLR_BOD5,C cross-sectional OLR (120 g BOD₅/m²/day) is much lower than 250 g BOD₅/m²/day, it means that the proposed design achieves the desired effluent quality, as well as it is much far from the risk of bed clogging.

This work proposes a practical and easy-to-apply method for sizing subsurface flow constructed wetlands (SSF CWs). It is applicable for all sources of greywater, especially those discharged from public baths. Crop ET is determined using the crop coefficient (K_c) method. Knowing that the works that have addressed the design problem of SSF CWs have not provided any analytical method to determine K_c. This work allows practitioners to quantify it by adopting the FAO 56 method that takes into account the type of plant, its growth period, the climatic conditions of the site, the media size and the frequency of wetting. The results obtained by the case study show the flexibility of the proposed methodology, such that the removal efficiency of TSS, BOD5 and COD was in accordance with the standards of irrigation waters (the desired application in this case). Therefore, the surfaces obtained from the SSF CWs will be compatible with the need for treatment avoiding any kind of oversizing of the system that implies larger surfaces or undersizing that causes treatment efficiencies that do not achieve the desired objective. This accounting to the needs of the treated water application allows to have a control of the design of the SSF CWs.

This work contributes to the progress of the use of decentralized treatment systems, especially CW. It encourages owners of public baths to implement this type of ecosystem, which provides ecological treatment and only requires local equipment that is easy to handle. We hope that this study will help local communities to more easily implement local water reuse programs for non-drinking purposes (irrigation, fire protection, toilet flushing and so on) and reduce the inappropriate demand for treated drinking water.

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

The authors declare there is no conflict.