Granular media filtration for on-site treatment of greywater: A review

Greywater, wastewater generated from different domestic sources except toilets, has been thought of as a reliable alternative to bridge the gap between increasing water demand and water supply (Al-Omari & Al-Houri 2021). Based on the degree of pollution, wastewater from hand basins, bathrooms and showers is classified as light greywater (LGW) while dark greywater (DGW) includes wastewater from the kitchen, dishwasher, laundry and washing machine. Use of raw greywater for landscape irrigation and reuse for non-potable purposes such as toilet flushing, garden irrigation and car washing is not safe (Garg et al. 2021; Glover et al. 2021; Patil et al. 2022). Significant variations in quantity and quality characteristics of greywater make the selection of greywater treatment process complicated (Kumar et al. 2022).

Several physical, chemical and biological treatment alternatives are adopted for the treatment of greywater, each of which has its own advantages and limitations. Chemical systems of greywater treatment use chemical coagulants and sorbents which may not be available locally. Large quantity of sludge generation is another issue which adds to the cost for further treatment (Vinitha et al. 2018). Biological treatment methods like sequential batch reactor (SBR), moving bed biofilm reactor (MBBR) and up-flow anaerobic sludge blanket (UASB) are proved to be good at the removal of different pollutants from greywater but they demand high energy and high operation and maintenance costs. SBR, RBC and MBR are less effective in the removal of phosphates from greywater (Kurniawan et al. 2021). Use of these biological methods at a decentralized level specifically in developing countries is not feasible. Further, biological systems for greywater treatment were reported to be sensitive to the variations in quantity and quality of influent greywater specifically at decentralized levels (Khalil & Liu 2021). Multiple studies recommended the treatment of source-separated greywater, while biodegradability of source-separated greywater is questionable (Noutsopoulos et al. 2018; Delhiraja & Philip 2020). Also, significantly less nutrient availability in greywater makes the use of biological systems for the treatment of greywater difficult (Zipf et al. 2016).

Natural systems such as constructed wetlands are energy and cost-efficient and do not require chemicals but they suffer from large land area requirements as high as 0.2–3.0 m²/person. Ramprasad et al. (2017) reported higher evapotranspiration rates in hot climates while Paulo et al. (2009) reported the problems of mosquitoes and flies in natural systems.

Physical systems are, in general, energy efficient and granular media filters are compact and are specifically suitable in urban areas where the availability of land is a great concern. Granular media filtration is one of those various treatment alternatives which has been used for a long time. The biggest advantage of the granular filtration technique is the use of locally available materials. A number of studies has been reported on greywater treatment using granular filtration. Different materials such as sand, lava rock and gravel have been used as filter media for treating greywater. Effects of hydraulic loading rate (HLR), organic loading rate (OLR), saturation conditions, depth of media and media size have been reported in the literature.

A few recent studies reviewed the feasibility and performance of membrane filters (Wu 2019), constructed wetlands (Arden & Ma 2018), oxidation and disinfection (Gassie & Englehardt 2017), green walls (Pradhan et al. 2019) and membrane bioreactors (Cecconet et al. 2019) for greywater treatment but no review on the use of granular media filters for greywater treatment has been reported. In the present study, an attempt has been made to summarize the recent literature available on the use of granular filters for greywater treatment. Constructed wetlands, green walls and green roofs without vegetation were also considered in the present study. Filter media characteristics and their effect on the filter performance are summarized. Effects of various parameters such as operating conditions, mode of operation, depth of filter media, media size and influent greywater characteristics on the performance of the filters are discussed. Different pre-treatments, which affect the performance of the filters are analyzed. An attempt has been made to suggest optimal granular filtration systems for the treatment of different streams of greywater. Finally, the need for further studies is highlighted.

The higher the weight of an item, the larger the label and the circle of an item (van Eck & Waltman 2020). The colour of an item is determined by the cluster to which the item belongs. In total four clusters were formed which are presented in four different colours, and all the clusters are closely related since the distance between two clusters denotes their relatedness. Days of filter operation and HLR are the repetitively used operational parameters while toilet flushing and landscape irrigation are the most frequently considered greywater reuse alternatives. Most of the studies considered the reduction of total suspended solids (TSS), turbidity, biochemical oxygen demand (BOD), surfactants, total coliforms and E. coli. Effect of treated greywater discharge on soil properties was also considered in multiple studies as indicated by frequent occurrence of the words soil, effect and risk.

Different locally available materials such as sand, gravel, slate, xylit, filtralite, Kanuma soil, oil shale ash, peat, charcoal, pumice, zeolite, zeolite tuff, blast furnace slag, loofah, teff straw, barks of pine, plant based biochar, indica and marmelos have been utilized as filter media in various studies (Itayama et al. 2006; Karabelnik et al. 2012; Ushijima et al. 2013; Assayed et al. 2014; Katukiza et al. 2014a; Dalahmeh et al. 2014a; Zipf et al. 2016; Al-Zou'by et al. 2017; Bernardes et al. 2017; Yaseen et al. 2019; Bahrami et al. 2020; Emslie et al. 2021; Prabha et al. 2021; Spychala et al. 2021). In addition, biochar, granulated activated carbon (GAC), bentonite clay, crushed lava rock and ceramic are a few other materials that are used as filter media. Different types of sand including river sand, quartz/silica sand, volcanic tuff sand, swimming pool sand, builders’ sand and dune sand have been used as a filter media by various researchers (Friedler et al. 2006; Chaillou et al. 2011; Assayed et al. 2015; Alsulaili et al. 2017; Bolton & Randall 2019). Generally, preference for locally available materials has been given in the literature to reduce the cost of the material. Non-availability and high cost of activated charcoal and GAC in rural areas make their use difficult though their performance for the treatment of greywater is better compared to other filter materials. Desired gradation or size of the filter media was obtained by sieving the media according to either ASTM (1998) or the CAWST filter manual (2012), (Assayed et al. 2014; Bolton & Randall 2019). Table 1 presents the characteristics of the filter media used in a few reported studies.

Washing the filter media before packing it into the filter makes it free from small particles. Kennedy et al. (2012) reported early clogging of unwashed filter media compared to that of washed media. Most of the reported studies washed filter media before packing into the filter column with demineralized water, tap water, potable water or hydrogen peroxide (Ghaitidak & Yadav 2016; Spychala et al. 2019; Shaikh & Ahammed 2021a).

Filter media of size ranging from 0.09 mm to >80 mm has been used in different studies (Bahrami et al. 2020; Sabogal-Paz et al. 2020). CAWST filter manual (2012) recommends the use of sand media with an effective size (D₁₀) of 0.15–0.20 mm for efficient working of the filter. However, effective size ranging between 0.12 and 2.85 mm had been adopted in the filtration studies with greywater. Filter media passing through 4.75 mm sieve size with an effective diameter of less than 0.65 mm had been used in most of the studies on greywater. Pore space in the media contributes to the removal of pollutants through microorganism growth and air and water flow (Emerick et al. 1997). Ball & Harold (1997), while treating wastewater by a filter, found an increase in pollutant removal as a result of increased surface area and a decrease in the size of media. On the other hand, the use of too fine media leads to early clogging and the development of the anaerobic condition in the filter. To avoid early clogging in the filter media, the use of uniformly graded media is recommended. Hence, most of the studies utilized filter media with a uniformity coefficient (UC) less than 4. Ideally UC for filtering material should be between 1.5 and 2.5 (CAWST 2012). The smaller the uniformity coefficient higher will be the total pore space and the infiltration rate which will ensure ample oxygenation.

The porosity of the media also impacts the removal of different pollutants during filtration process. The filter media with porosity in the range of 28% and 85% have been utilized in various studies (Dalahmeh et al. 2014a; Subramanian et al. 2020). Specific surface area (SSA) of the media is a measure of surface area available for attachment of the pollutants and microbes. Generally, a higher SSA is associated with better removal of pollutants in a filter (Dalahmeh et al. 2014b).

The performance of filters is affected by various parameters such as loading rate, media characteristics, media depth, influent characteristics and mode of operation. Laboratory and pilot-scale filters were constructed using non-corrosive materials such as acrylic, plexiglass, poly vinyl chloride (PVC), unplasticised PVC (uPVC), glass, concrete and plastic (Al-Hamaiedeh & Bino 2010; Antonopoulou et al. 2013; Babaei et al. 2019; Sabogal-Paz et al. 2020). Most of the laboratory studies preferred cylindrical filters since rectangular or square shape filters have dead pockets at the corners while a few pilot scale studies made use of rectangular concrete filters (Patil & Munavalli 2016).

Hydraulic and organic loading rates are important parameters influencing the performance of the filters. Wide variations in the quality of filtrate with variation in hydraulic loading rate (HLR) and organic loading rate (OLR) have been reported in the literature. Filter performance was evaluated for a wide range of HLR (3.2 cm/day-42.1 m/h) and OLR (7.83–496.00 gBOD/m²/day) (Abdel-Shafy et al. 2014; Dalahmeh et al. 2014b; Katukiza et al. 2014a; Al-Zou'by et al. 2017). The selection of optimum HLR and OLR is important as they affect clogging and effluent quality. Both HLR and OLR greatly influence the days of filter operation and the quality of effluent. While Friedler et al. (2006) could operate the filter only for 15 min at an HLR of 8.28 m/h, Ghaitidak & Yadav (2016) operated their filter for 200 days at a low HLR of 7.02 cm/day.

Table 2 presents a summary of a few studies on the effect of HLR and OLR on the removal efficiency of different parameters in the filters. Increased HLR and OLR significantly reduced the potential of filters to remove organic content, nutrients and microorganisms. Assayed et al. (2014) reported no significant effect on turbidity removal when HLR and OLR were doubled. Similarly, in a study conducted by Patil & Munavalli (2016) turbidity removal was not influenced in sand filters when OLR was increased from 4 to 35 g COD/m²/day.

TSS removal in the filters is also least dependent on variations in HLR and OLR. For example, Ghaitidak & Yadav (2016) treated light greywater generated from a student hostel and reported TSS removal of 56% and 53% at an OLR of 89 and 153 g BOD/m²/day, respectively. Similarly, Assayed et al. (2014) reported TSS removal of 99.6% and 94.4% at HLR of 72 and 142 L/m²/day, respectively in the treatment of synthetic greywater. Unlike organic content and microorganisms, turbidity and TSS are removed by physical processes such as sedimentation, straining and adsorption. The first few days of filter operation leads to the development of a fine layer on the top of filter media which will act as a straining mat and contribute to the removal of solids even under increased loading rates (Katukiza et al. 2014a). Zipf et al. (2016) and Karabelnik et al. (2012) reported that variations in HLR and OLR did not significantly affect the pH and EC values of effluent greywater during filtration.

Removals of organic content, nutrients and microorganisms decreased in the filters when HLR and OLR were increased. BOD removal was reduced from 85% to 78% when HLR was increased 2.5 times in the treatment of mixed greywater from a household in Norway (Karabelnik et al. 2012). Katukiza et al. (2014a) reported a reduction in NH₄-N and total Kjeldahl nitrogen (TKN) removals from 69% to 52% and from 51% to 42%, respectively when HLR was increased from 20 to 40 cm/day. Increased filtration velocity at higher HLR provides shorter residence time to the greywater in the filter. Dalahmeh et al. (2014a) reported a decrease in phosphorus removal under increased HLR conditions.

Total coliforms and E. coli removal in a filter with crushed lava rock media was reduced to 1.62 from 1.83 log and to 2.31 log from 2.52 log when HLR was increased to 40 from 20 cm/day, respectively, in the treatment of dark greywater from households in Uganda (Katukiza et al. 2014a). Mandal et al. (2011) while treating light greywater from an Indian household reported an increase in TC and FC removal from 1.37 and 1.46 to 1.63 and 2.25 log when OLR was dropped from 15 to 8.4 gCOD/m²/day, respectively. Removal of organic matter, nutrients and microbiological parameters are influenced by microbial activity in the filters in addition to physical processes. Contact time between greywater and microbes decreases with the increase in HLR, which results in reduced removal. The susceptibility of the microbes to the variations in OLR is well known (Munavalli et al. 2022).

Bahrami et al. (2020) studied the potential of three different filter media namely zeolite, pumice and GAC at various contact times in a batch process while treating MGW from a dormitory in Iran. They reported 4 h as optimum contact time between greywater and the filter media to obtain maximum COD removal while further increase in contact time did not improve the COD removal by more than 5%.

Different techniques are used for filter media characterisation. Cation exchange capacity (CEC) indicates the capacity of positively charged cations in media to adsorb negatively charged particles such as clay and organic matter (Patel et al. 2022). Fourier transform infrared spectroscopy (FTIR) analysis of the media has been used in a few studies to analyse various functional groups present on the media (Patel et al. 2020). Surface area and pore volume are assessed using Brunauer-Emmett-Teller (BET) analysis. Morphological characteristics of the media are measured using scanning electron microscopy (SEM) in several studies (Babaei et al. 2019; Bahrami et al. 2020). Surface structures such as smoothness, undulations and roughness can be visualized using SEM analysis. Energy dispersive X-ray (EDX) analysis is used to determine the elemental composition of the media (Katukiza et al. 2014b; Patel et al. 2022). The chemical composition of the media can be measured using spectroscopic techniques such as X-ray fluorescence (XRF), Raman spectroscopy, X-ray photoelectron spectroscopy (XPS) and X-ray diffraction (XRD) (Chan & Radin Mohamed 2013; Katukiza et al. 2014b; Ramadan 2015).

Table 3 presents the summary of studies on the effect of media size on the performance of the filters treating greywater. Media size, specific gravity, porosity, permeability and SSA are some of the important media characteristics which influence the filter performance substantially. Increased media size deteriorates the filter performance in terms of solids, organic content, nutrient and microbial removal. Singh et al. (2021) reported a reduction in turbidity removals from 67% to 58% when media size was increased from 0.40–0.80 mm to 0.80–1.18 mm. Xu et al. (2020) also reported a significant reduction in turbidity removal when the size of the media was increased from 0.5–1.0 mm to 4.0–8.0 mm in the treatment of synthetic greywater. Abdel-Shafy et al. (2014) reported 82% and 54% removal of TSS while treating greywater by sand and gravel media, respectively, under similar operating conditions. Khalfan Al Jahmani et al. (2021) compared the performance of ceramic filters of particle sizes 0.25, 0.72 and 1.18 mm and found the highest TOC, COD, TSS and turbidity removals with 0.25 mm size while treating kitchen greywater from a household in Oman. This is mainly because of the increased attachment of contaminants in filters with smaller media sizes compared to filters with larger media (Ochoa et al. 2015).

The change of media size do not generally affect the pH of greywater during filtration. However, free hydrogen ions in the filter media can affect the pH of treated greywater (Xu et al. 2020). The higher ion-exchange capacity of the filter materials might contribute to an increase in the pH of effluent (Aghakhani et al. 2012). The rough and porous surface of the media facilitates ion exchange, and as a result, effluent pH remains unchanged. An increase in media size significantly reduces organic content removal efficiency in the filters (Bahrami et al. 2020). Abdel-Shafy et al. (2014) reported a significant reduction in BOD and COD removals from 76% to 46% and 74% to 41%, respectively, when sand media with 1–2 mm size was replaced by gravel media of 2–4 mm. A study by Singh et al. (2021) also observed the negative effect of increased media size on the organic content removal efficiency of the filters.

Nitrogen removal is also influenced by media size of the filter. For example, Ochoa et al. (2015) during the treatment of light greywater from Japanese households reported 75% and 53% removal of total nitrogen with a media of effective sizes 0.3 and 0.6 mm, respectively. Abdel-Shafy et al. (2014) reported NH₄-N removal of 40% and 55% when gravel and sand media were used to treat mixed greywater from an Egyptian household. Torrens et al. (2009) found nitrification, denitrification and total nitrogen removal efficiencies were inversely proportional to the square root of sand size. Singh et al. (2021) evaluated the performance of sand filters with media size 0.40–0.80 mm and reported the TC and FC removals of 1.93 and 1.66 log, respectively, which were reduced to 1.60 and 1.36 log when the media size was increased to 0.80–1.18 mm.

Smaller size filter media with a higher SSA provide additional surface for the attachment of microbes and contaminants resulting in better removal of contaminants. Dalahmeh et al. (2014b) treated synthetic greywater and reported 61% and 92% removal of TSS using sand and pine bark which had SSA of 0.136 and 0.734 m²/g respectively while BOD removal was 75% and 98% in the sand and pine bark filter, respectively. The difference in hydrophobicity of the media could be responsible for variation in the treatment potential of different filter media (Bahrami et al. 2020). Filter media such as GAC with higher SSA, permeability and porosity have good adsorption capacity and contribute to significant removal of pollutants (Zipf et al. 2016). The bulk density of the filter media decides the mass of the media that can be occupied in the filter column. The treatment efficiency by a specific filter media is dependent on the mass of the media in the filter column. For example, Bahrami et al. (2020) while treating mixed greywater (MGW) reported that pumice was more efficient than zeolite when equal mass of both of them was considered. In other words, the bulk density of the filter media decides the contact time between greywater and the filter media.

Derjaguin-Landau-Verwey-Overbeek (DLVO) theory is used to explain the interactions of colloids with media surface (Derjaguin 1941; Verwey & Overbeek 1955). According to DLVO theory, interaction energy can be attractive or repulsive and can be quantified as the sum of van der Waals and electrostatic double-layer interactions (da Silva et al. 2011; Sadeghi et al. 2013; Torkzaban et al. 2013). The strength of interactions depends on several factors such as temperature, pH, ionic strength, metal oxides, presence of organic matter and multivalent ions like calcium (Redman et al. 2004; Kim & Walker 2009; Furiga et al. 2010; Bradford et al. 2014; Sasidharan et al. 2016). Surface roughness and chemical heterogeneity of media and colloidal surface play an important role in the interaction energy between a colloid and collector (Bradford & Torkzaban 2013, 2015).

Development and maturation of biolayer in the filters is a long-term process and physicochemical processes such as sedimentation, straining, diffusion and inertial and centrifugal forces contribute to improving the quality of greywater when the biolayer is not fully mature. Media properties such as particle size, porosity and roughness decide the removal of colloids by these mechanisms (Sasidharan et al. 2016). Size and concentration of colloids along with hydrodynamic conditions and solution chemistry play an important role in colloidal retention by straining and attachment (Bradford & Torkzaban 2013, 2015). Larger particles are removed by straining while colloidal particles are removed by attractive forces (Song et al. 2020). Straining, adsorption, interception and diffusion are principal mechanisms responsible for removal of bacteria (Weber-Shirk & Chan 2007; Schijven et al. 2014; Napotnik et al. 2021). Contaminants such as nitrogen, phosphorus, potassium, chloride, aluminium and enzymes along with microorganisms can also be trapped since sorption is not specific with the compound (Flemming et al. 2016; Sabogal-Paz et al. 2020).

Sasidharan et al. (2016) evaluated the potential of quartz sand to retain three viruses namely MS2, PRD1 and ɸX174, and E. coli, and reported significant removal by attachment. In other studies, Bradford et al. (2015a) and Bradford et al. (2015b) reported negligible attachment of E. coli to Ottawa sand surface. Metal oxides and clay particles on sand surface increase microbial removal by enhancing the number of favourable sites for attachment (Tong et al. 2012). Mandal et al. (2011) evaluated the performance of filter in the removal of Salmonella spp. while Dalahmeh et al. (2012) studied the removal of thermotolerant faecal coliforms in the filters. Both the authors reported more than 90% removal of these microbes during filtration.

Most of the studies on granular media filtration used single media while a few recent studies evaluated the potential of multimedia filters (Mahmoudi et al. 2021; Prabha et al. 2021). Table S1 (Supplementary material) presents comparison of single media and multimedia filters in removal of different parameters. As can be seen from Table S1, combining two media did not significantly improve the filter performance in the removal of different contaminants. Amiri et al. (2019) studied the potential of zeolite alone and its combination with GAC to treat MGW from a student hostel in Iran. They reported COD removal of 68% with zeolite which increased to 73% when the combined system was adopted. Xu et al. (2020) also reported that the use of combined media instead of single media did not improve the effluent quality. Abdel-Shafy et al. (2014) evaluated the potential of gravel alone and its combination with sand to treat MGW from an Egyptian household and reported no significant difference in the performance. The single or multimedia filters operated in unsaturated conditions did not alter the pH of influent greywater (Xu et al. 2020). However, a detailed study comparing the potential of single and multimedia filters to treat greywater is needed. The use of coarse media at the top layer would be beneficial in increasing the life of the filter without compromising the quality of effluent.

Al-Zou'by et al. (2017) used silica sand, volcanic tuff sand and zeolite with porosity of 30%, 38% and 41% for the treatment of mixed greywater and reported the highest removal by sand compared to the other two media. Activated charcoal and GAC have good adsorption and ion exchange capacity and high surface area and porosity leading to their use as filter media (Dalahmeh et al. 2014b; Zipf et al. 2016). For example, Dalahmeh et al. (2014a) reported surface media of sand as 0.14 m²/g compared to >1,000 m²/g for activated charcoal. A few studies used activated charcoal and GAC as filter media either separately or in combination with other filter materials to improve the performance of the filters (Assayed et al. 2014; Zipf et al. 2016). Bahrami et al. (2020) while treating synthetic greywater reported 50%, 30% and 10% removal of COD in filters with GAC, zeolite and pumice media. Filters with silica sand, volcanic tuff sand and zeolite tuff as filter media reported 89%, 58% and 77% of turbidity removal, respectively (Al-Zou'by et al. 2017). GAC filter media showed about 5–25% higher removal of organic matter compared to fine sand filter because of the significantly higher surface area of GAC (Dalahmeh et al. 2014b).

Zeta potential (ZP) is a measure of electrophoretic mobility of particles and indicates the stability of a colloidal system (Bodnar et al. 2014). Whether particles will remain in suspension by repelling each other or settle by agglomeration is indicated by their ZP (Priyatharishini & Mokhtar 2020). The surface charge density of a microorganism can be calculated from its ZP (Khalaphallah & Andres 2012). pH has a significant effect on ZP with ZP value being close to zero when the pH of the greywater is near neutral (Barisci & Turkay 2015). High negative ZP was reported by Bodnar et al. (2014) for laundry and kitchen greywater samples which have higher turbidity and EC values compared to bathroom greywater indicating a positive correlation of ZP with turbidity and EC values of the greywater samples. This is due to the presence of large colloidal solids in kitchen and laundry greywater compared to bathroom greywater.

Media depth is an important parameter that can affect filter performance significantly. Ellis & Wood (1985) recommends a minimum depth of 30 cm for filter media for efficient removal of different physicochemical and microbial parameters and hence, most studies utilized a minimum media depth of 30 cm (Aizenchtadt et al. 2009; Zipf et al. 2016; Thompson et al. 2020; Sharaf & Liu 2021). However, a few studies while assessing the feasibility of compact systems used media depth of less than 30 cm (Griffioen & Natha 2013; Ward et al. 2015; Prabha et al. 2021). Removal of most of the contaminants occurs in the top 10–30 cm depth of filter media. Most studies on treating greywater utilized media depth in the range of 30–100 cm as occurrence of anaerobic conditions at greater depths has been reported (Katukiza et al. 2014a; Shaikh & Ahammed 2021b).

Bacterial activity is dominant at the upper layer of the filter and reduces along with the depth of the filter due to a decrease in oxygen supply and food in terms of organic matter (Ranjan & Prem 2018). The formation of the biolayer at the top of the media helps in the significant reduction of turbidity by the top layers (Katukiza et al. 2014b). Most of the turbidity removal occurred in the top 30 cm depth in the filters while the additional filter depth contributed little to turbidity removal (Verma et al. 2019). Sedimentation, precipitation and straining are major mechanisms of turbidity removal in granular filters (Dalahmeh et al. 2014a). Slight to negligible variation in pH along the depth of filter media has been reported (Shaikh & Ahammed 2021b). Most of the studies reported an increase in the EC value of influent greywater with an increase in media depth (Bahrami et al. 2020; Emslie et al. 2021). A slight increase in pH and EC of influent greywater was attributed to the phenomenon of filter media leaching and free hydrogen ion concentration in the filter media (Spychala et al. 2019; Xu et al. 2020; Spychała et al. 2021).

Most of the studies reported reduction in dissolved oxygen (DO) levels during the filtration process (Xu et al. 2020; Singh et al. 2021). Sufficient availability of oxygen to the aerobic heterotrophic bacteria promotes the removal of organic content by aerobic oxidation (Abed et al. 2017; Cheng et al. 2021). About 50% of influent BOD and COD removals occurred in the top 10 cm depth of the filter media while the remaining depth contributed additional removals (Katukiza et al. 2014a). Ochoa et al. (2015) reported a significant reduction in influent COD concentration in the top 10 cm media depth while treating greywater from Japanese households. Biodegradation and mineralization by diverse microbial communities are principal mechanisms responsible for the removal of organic matter in the filters (Khalil & Liu 2021). Optimum depth of media depends on several factors such as quality and quantity of influent greywater, mode of operation and media characteristics. Shaikh & Ahammed (2021a) reported 52% removal of COD in the top 10 cm depth of the media while 50 cm was the optimal depth of fine sand (0.30–1.18 mm) at an HLR of 32 cm/day for the removal of organic content in the treatment of mixed greywater. Katukiza et al. (2014a) reported 50% COD removal in the top 10 cm depth of the media (1.18–2.36 mm) at an HLR of 20 cm/day in the treatment of dark greywater from households in Uganda. Based on the reported studies the optimum depth for the removal of organic content with fine sand media (media size less than 2.36 mm) is 50 cm when operated at HLRs of less than 32 cm/day.

Depth of sand bed is more important in nutrient removal as the occurrence of simultaneous nitrification-denitrification along with the filter depth has been reported (Murphy et al. 2010). Top 10 cm reduced the majority of the NH₄-N from the greywater while the rest of the media depth helped improve the NH₄-N removal. Similarly, though most of the PO₄-P removal occurred in the top layer, the removal improved along with the depth of filter media.

The presence of biolayer at the top of filter media enables higher microbial removal in the top 10–30 cm depth of the filter media. Verma et al. (2017) reported 67% removal of TC in the top 10 cm depth. The presence of biolayer at the top of filter media after the initial few days of operation was reported by Emslie et al. (2021). Other studies did not evaluate the potential of filters to remove microbiological parameter along the depth.

Figure 2(c) and 2(d) shows that percentage COD and BOD removals in the filter are sensitive to the quality of influent greywater with higher removal at higher influent concentrations. This might be because of the higher oxidation rate of organic matter by microbes in greywater with high influent organic content. However, hydrophobicity in the filter media particles may affect the performance of the filters (Maimon et al. 2017). For example, Comino et al. (2013) reported 84% COD removal when influent COD was 234 mg/L compared to 76% for an influent COD value of 77 mg/L. The ratio of COD/BOD is a measure of biodegradability of greywater and a COD/BOD ratio of less than 2 indicates good biodegradability. Biodegradability of greywater from different sources varies significantly, with kitchen greywater showing highest biodegradability (Shaikh & Ahammed 2020). A higher reduction of BOD in the filter indicates a healthy microbial population while the reduction in COD could also be due to adsorption onto the filter media (Prabha et al. 2021). However, several studies observed lower COD removal compared to BOD removal, indicating the presence of slowly and non-biodegradable organic matter in the greywater (Aizenchtadt et al. 2009; Katukiza et al. 2014b). Adsorption and biodegradation are major mechanisms responsible for the removal of BOD and COD in the filters (Assayed et al. 2015).

The nutrient removal efficiency of the filter increases with an increase in influent nutrient concentration (Figure 2(e)). The majority of the nitrogen removal in the filters occurs due to the adsorption process which is dependent upon the pH of the influent greywater (Katukiza et al. 2014b). Detergents, soaps and traces of urine are major sources of PO₄-P in greywater (Mohamed et al. 2018; Noutsopoulos et al. 2018). A pH range of 5.5–8.0 in.the influent greywater favors the process of nitrification specifically in saturated filters as wet sand plays an important role in the nitrification process (Tonon et al. 2015). Dallas & Ho (2005) reported 2.46 log removal of FC while treating mixed greywater in subsurface flow reedbeds without vegetation when influent concentration was 8.5 × 10⁷ CFU/100 mL.

The concentration of microbial parameters in raw greywater influences the degree of microbial removal during filtration (Figure 2(f)). Total coliforms, faecal coliforms and other parameters are used as microbial indicators. The presence of these microbial parameters in influent greywater was detected in all the studies conducted on the filtration systems treating greywater.

Sabogal-Paz et al. (2020) reported better turbidity removal in continuous filters compared to intermittent filters. Shaikh & Ahammed (2021a) reported 71% and 87% removals of BOD and COD, respectively, in the continuous filter which were reduced to 59% and 79% in an intermittent filter. Continuous feeding allows longer retention time in continuous filters, resulting in better removal of physicochemical and microbial pollutants compared to intermittent filters. Shaikh & Ahammed (2021a) reported significantly better performance of continuous filters in NH₄-N and PO₄-P removal compared to intermittent filters. Higher E. coli removal was also reported when filters were operated in continuous mode compared to intermittent operation (Verma et al. 2017). The effect of increased HLR and OLR was more pronounced in intermittent filters compared to continuously operated filters (Shaikh & Ahammed 2021a).

In intermittent filters, the frequency of dosing is an important parameter. Wide variations in the daily dosing ranging from a single dose to 40 doses per day have been reported for intermittent filters in the literature. Dalahmeh et al. (2014a) fed 70%, 10% and 20% of total daily greywater volume at 9, 16 and 20 h, respectively. Filters were fed 8 and 11 times a day by Assayed et al. (2014) and Friedler et al. (2005), respectively. Media uniformity becomes more significant with less frequent dosing as pore geometry and condition under which thin-film flow occurs get affected. Resting of the water to be treated for the duration of 1–48 h between two consecutive feedings (pause period) decides the efficiency of intermittent filters (Sabogal-Paz et al. 2020). Continuously operated filters need control over filtration rate and therefore, might need a higher area for installation if need to be operated at a higher filtration rate (Maciel & Sabogal-Paz 2020).

Shaikh & Ahammed (2021a) reported that turbidity removal is independent of the saturation condition of the filters. In unsaturated filters, the pH and EC were not altered indicating that unsaturated filters neither absorb nor releases any ions (Chaillou et al. 2011; Prabha et al. 2021) while a slight increase in the pH and EC were reported by a few researchers who used saturated filters (Karabelnik et al. 2012; Antonopoulou et al. 2013). Saturated filters are more efficient in the removal of BOD and COD from greywater compared to unsaturated filters. Higher BOD reduction in saturated filter compared to unsaturated filter could be due to degradation process by microorganisms adhered to the media surface (Young-Rojanschi & Madramootoo 2014).

Higher removal of PO₄-P has been reported in saturated filters compared to unsaturated filters (Shaikh & Ahammed 2021a). This might be because of the lack of electrostatic force of attraction in unsaturated filters (Kaewsarn & Yu 2001). Undisturbed slime layer in saturated filters favours PO₄-P removal from greywater as a result of biodegradation and adsorption (Abdel-Shafy et al. 2014). On the other hand, presence of aerobic condition in unsaturated filters favors higher NH₄-N removal compared to saturated filters which have anoxic conditions at greater depths (Achak et al. 2009). Adsorption and biodegradation are major mechanisms that reduce influent PO₄-P concentration in the filters (Assayed et al. 2015), and these mechanisms are dependent upon the saturation condition of the filters.

The presence of undisturbed biolayer in saturated filters helps in better removal of microbial parameters in saturated filters compared to unsaturated filters. Formation of the biolayer at the top of the media results in the higher reduction of flow rate in saturated filters compared to unsaturated and that contributes to additional microbial removal in saturated filters compared to that in unsaturated filters (Shaikh & Ahammed 2021a). The effect of increased HLR and OLR was more pronounced in unsaturated filters compared to that in saturated filters (Shaikh & Ahammed 2021a). Shaikh & Ahammed (2021a) reported that turbidity removal was independent of the pause period while the effect of pause period was significant in saturated filters compared to unsaturated filters in the removal of organic content, nutrients and microbial parameters.

Filters can be operated in parallel to treat the higher quantity of greywater (Katukiza et al. 2014a) or can be operated in series to improve the quality of effluent (Subramanian et al. 2020). Most of the studies utilized single filters compared to the use of multiple filters either in parallel or in series. Vertically operated filters are more efficient in removing different pollutants from greywater compared to horizontal filters (Karabelnik et al. 2012). Downflow vertical filters are not only energy-efficient but also have a higher potential to remove the majority of the pollutant compared to upflow filters (Abdel-Shafy et al. 2014). Patel et al. (2020) and Abdel-Shafy et al. (2014) evaluated the performance of the upflow filters while Mandal et al. (2011) and Patil & Munavalli (2016) studied combined upflow-downflow filters for the treatment of greywater.

Filters treating greywater may face many operational troubles such as early clogging and short-circuiting. Quantity and quality of influent, mode of filter operation, media characteristics such as size and depth and pre-treatment are among the factors that influence the run time of the filters.

Early clogging of the filters is among the most influencing factors that restrict the use of filters for the on-site treatment of greywater. Clogging results in head loss development in the continuously operated filters while it causes a reduction in the flow rate in intermittent filters. The rate of head loss development in the filter is directly related to the feed volume and the media size. Limited data are reported in the literature on head loss development and flow rate reduction in filters. Bahrami et al. (2020) stopped the continuous feeding to the filter once the head loss reached 30 cm while Katukiza et al. (2014b) reported a reduction in flow rate from 1.1 to 0.5 L/min after 30 days of filter operation. Small impurities get trapped into the empty pores of the filters with use and a thin biologically active layer is formed at the top of the filter media resulting in reduced flow rate with time.

To reduce clogging and to increase the length of filter run different strategies have been used by the researchers. A few studies considered pre-treatment of greywater while others modified the filters to reduce early clogging in the filters. Assayed et al. (2014) adopted a drawer compacted sand filter with multiple identical drawers which can be taken off in case of clogging or for maintenance. Ochoa et al. (2015) made use of geotextiles before filtration to remove fine particles from influent greywater. The size of the filter media adopted also decides the rate of clogging in the filters with occurrence of early clogging in filters with fine media compared to coarser media (Bahrami et al. 2020). A few researchers adopted coarser media at the top followed by fine media at the bottom (Kwabena Ntibrey et al. 2020). Separation of the different media layers by geotextile could prevent the downward movement of media (Moges et al. 2017).

Early clogging in the filters can be avoided by reducing the suspended solid concentration of the influent greywater by means of pre-treatment. Pre-treatments such as sedimentation, screening, equalization, coagulation and coarse filtration have been used either separately or in combination (Friedler & Alfiya 2010; Antonopoulou et al. 2013; Ghaitidak & Yadav 2016; Zipf et al. 2016). Ghaitidak & Yadav (2016) adopted a coarse filter as pre-treatment while Patil & Munavalli (2016) adopted screening and sedimentation as pre-treatment. Mandal et al. (2011) combined screening, coarse filtration and equalization as pre-treatment steps to the filtration system. The selection of the pre-treatment method is dependent upon characteristics of influent greywater, targeted reuse and the main treatment option adopted.

Though both sources contribute almost equally to the total greywater volume generated from households, large variability in the concentration of different pollutants in LGW and DGW sources had been reported in the literature (Noutsopoulos et al. 2018). Hence, source-based treatment of greywater is recommended in multiple studies (Katukiza et al. 2015; Noutsopoulos et al. 2018; Oteng-Peprah et al. 2018). LGW sources are less turbid compared to DGW sources and hence pre-treatment such as settling, screening and coarse filtration reduces the solid concentration to a great extent while for kitchen and laundry greywater pre-treatments such as chemical coagulation, constructed wetlands and coarse filtration can be used based on cost consideration and land availability. Coagulation (Vinitha et al. 2018), CWs (Ramprasad et al. 2017), and coarse filtration (Ghaitidak & Yadav 2016) were found to be efficient for the removal of solids from greywater.

Filter cleaning is done by cleaning the top few centimetres of the media (Katukiza et al. 2014b) or by replacing this with new media (AL-Zou'by et al. 2017). A few researchers adopted backwashing before resuming the filter operation once flow rate reduced substantially in intermittent filters or significant head loss occurred in continuously operated filters (Babaei et al. 2019). It should be noted that resetting of the filter either by backwashing or scouring the filter surface could be performed multiple times, but these techniques would not work after a few resettings as filter media got heavily clogged and media may require replacement (Freitas et al. 2021). A reduction in flow rate after resetting the filter has been reported in the literature (Singh et al. 2021).

Early clogging in the filter could also be reduced by using uniformly graded media. The smaller the uniformity coefficient the higher will be the pore space and infiltration rate which will ensure ample oxygenation (Dalahmeh et al. 2014a). By selecting the proper mode of operation, early clogging can be restricted to a great extent as Shaikh & Ahammed (2021b) reported a difference in flow rate reduction for saturated and unsaturated filters. Use of a saturated filter for LGW and MGW treatment, for example, could be a better alternative compared to that for DGW. The size of the filter media plays a crucial role in deciding the rate of clogging in the filter. Fine sand media resulted in early clogging compared to coarse media but the use of coarse media resulted in poor performance of the filters (Xu et al. 2020). Hence, the use of coarse media at the top followed by fine media could be used to prevent early clogging in the filter specifically for the saturated filters.

Short-circuiting in the filters could affect the performance of the filters in reducing different pollutants. Tracer tests were performed by a few researchers to understand the deviation from ideal plug flow conditions due to dispersion in the flow paths through porous media. Morrill Dispersion Index (MDI) is a measure of flow conditions through the filter media. As stated by USEPA (1986) and Tchobanoglous et al. (2011), the MDI characterizes all the filters as plug flow reactors when the MDI value is less than 2. In a plug flow reactor, the fluid passes through the filter with no mixing of earlier and later entering fluid (no overtaking). The necessary and sufficient condition for plug flow condition is that the residence time in the reactor must be the same for all elements of the fluid (Levenspiel 1999). Sabogal-Paz et al. (2020) reported an MDI value of 1.54 which was lower than the value of 1.8 reported by Young-Rojanschi & Madramootoo (2015). Therefore, from the perspective of the biological layer development and microbial removal processes, the results suggest the similar time is available for all portions of water that enter the filters at the respective flow rate, helping the greywater treatment.

It should be noted that different reuse applications have different standards or guidelines and the selection of treatment technology depends on the reuse purpose. Most of the guidelines consider solids (turbidity and TSS), organic content (BOD and COD) and bacteriological indicators (total coliforms, faecal coliforms and E. coli). Some countries like China have considered other parameters such as nitrogen and phosphorus in their standards for wastewater recycling (Boyjoo et al. 2013). World Health Organization (WHO 2016) has released restricted and non-restricted reuse guidelines in 2016 which are more focused on microbial quality compared to physical and chemical quality. Counties such as Australia, Japan and the USA have pre-defined greywater reuse guidelines as they are already recycling and reusing the greywater while countries like Jordan and India have less rigorous standards for greywater reuse compared to the developed countries (Vuppaladadiyam et al. 2019).

Table S3 (supplementary material) summarises the reported performance of filters reported in the literature. Most of the studies met the USEPA (2012) guideline for TSS of 30 mg/L for restricted access area irrigation and construction (Figure 4). Because of the higher initial turbidity and low efficiency of filtration, only a few studies met the USEPA (2012) standard of 2 NTU for urban and agricultural reuse. Almost all the studies reported effluent pH values in the range of 5–9 meeting the USEPA (2012) standards for restricted access area irrigation, urban, agricultural and indirect potable reuse. Filters operated at a low loading rate with fine media and low influent concentration met the WHO (2016) BOD standard of 10 mg/L for toilet flushing while filters operated at moderate loading rate with fine to coarse media met the USEPA (2012) standard of 30 mg/L for all the purposes.

Limited data is available on the removal of pathogens in the filters even after many years of use of filtration systems for greywater. Sabogal-Paz et al. (2020) could meet the WHO (2016) standard for unrestricted irrigation of crops by adopting microbial fuel cells as a pre-treatment to the saturated sand filter while Mandal et al. (2011) could meet the WHO (2016) standard of 10⁵/100 mL for restricted irrigation while treating MGW. The rest of the reported studies failed to meet either of the standards mainly because of the use of coarser media and higher loading rates (Katukiza et al. 2014a; Ghaitidak & Yadav 2016; Zipf et al. 2016).

Several studies reported the inability of filtration systems to meet different reuse standards worldwide. This indicates the need for proper pre-treatment and post-treatment to the filtration system. Significant variations in the quantity and quality of greywater among the different sources of greywater have been reported. LGW is less polluted compared to DGW, though both contribute equally to the total greywater volume generated from households (Shaikh & Ahammed 2022). Hence, separate treatment of LGW and DGW is recommended by multiple studies (Katukiza et al. 2015).

Pre-treatments like screening, settling, aeration, coagulation, equalization, coagulation, coarse filtration, geotextile mesh, constructed wetlands and microbial fuel cells have been adopted either individually or in combination (Mandal et al. 2011; Ghaitidak & Yadav 2016; Sabogal-Paz et al. 2020). Screening, settling, equalization and coagulation are most frequently used pre-treatment alternatives (Sostar-Turk et al. 2005; Patil & Munavalli 2016; Zipf et al. 2016; Alsulaili et al. 2017). Post-treatments such as disinfection and GAC adsorption are most commonly used (Al-Ismaili et al. 2017; Noutsopoulos et al. 2018; Prabha et al. 2021). Membrane filtration, charcoal filter, aeration and constructed wetlands were also used as post-treatment alternatives to achieve different reuse standards (Mandal et al. 2011; Patil & Munavalli 2016; Babaei et al. 2019).

Greywater from the hand basin, bathroom and laundry are less biodegradable compared to greywater from the kitchen (Khalil & Liu 2021). Light greywater sources have significantly less organic content compared to DGW (Dwumfour-Asare et al. 2018; Babaei et al. 2019). Pre-treatment such as sedimentation and coarse filtration along with granular filters with appropriate media characteristics and operating conditions could reduce the organic content in the influent greywater substantially to meet the different reuse standards. Use of a coarse media layer on the top of the filter is recommended in both saturated and unsaturated filters as it could prevent the early clogging in the filter and increase the life of the filter (Babaei et al. 2019). Hence, use of continuous filter over intermittent filter whenever possible is recommended as the former is more efficient in the removal of different pollutants than the latter.

Light greywater can be subjected to sedimentation and equalization up to 12 h or at least for 1 h to remove settleable solids. After settling, the use of a saturated filter with fine media or GAC of a minimum 50 cm depth is recommended to meet the restricted reuse standards for non-potable purposes (Shaikh et al. 2019). The choice of GAC over fine size media depends upon the quality of greywater after pre-treatment and reuse requirements. The greater media depth helps in the reduction of nitrogen from greywater as the occurrence of simultaneous nitrification-denitrification has been reported in saturated filters with greater depths (Achak et al. 2009). A minimum depth of 50 cm is recommended for filters as removal of organic content, phosphorus and bacterial load has been reported throughout the depth of the media though the majority of these pollutants are removed in the top 10 cm depth (Katukiza et al. 2014b). Filtered greywater could be disinfected to meet the unrestricted reuse standards if needed.

Dark greywater sources have a significantly higher concentration of solids, organic content, organic micropollutants, nutrients and heavy metals compared to LGW sources (Khalil & Liu 2021). Hence, pre-treatments like coagulation and constructed wetlands are recommended to reduce load of these pollutants substantially. Coagulation could remove 70–80% of solids, 60–70% of organic content and 40–50% of nutrients along with 1–3 log reduction in bacterial load in influent greywater (Pidou et al. 2008; Ghaitidak & Yadav 2015; Vinitha et al. 2018). Disinfection by chlorine or sodium hypochlorite can be done post-filtration based on the filtrate quality to meet the unrestricted reuse standards. Constructed wetlands could be used as a pre-treatment in rural areas while coagulation could be used in urban areas. Constructed wetlands can remove 50–60% of suspended solids and organic content and 80–90% of the microbial load from influent greywater by physicochemical and biological processes. Unsaturated filters can be used post-CWs as a large fraction of solids, organic content and microbial pollutants will be removed by CWs. An unsaturated filter with fine to medium size media with a minimum 50 cm depth can be used. Based on the quality of filtrate, the use of GAC or disinfection is recommended to meet the different unrestricted reuse standards.

The biolayer present at the top of filter media plays a significant role in the removal of different pollutants including organic content and microorganisms from influent greywater. A detailed study needs to be conducted to evaluate the effect of nutrient concentration on the rate of formation of biolayer. Also, the factors that affect biolayer formation need to be studied in detail for different greywater sources as the difference in the level of biodegradability of different greywater sources has been reported in several studies (Noutsopoulos et al. 2018; Oteng-Peprah et al. 2018). Several techniques such as scanning electron microscope (SEM) with energy-dispersive X-ray spectroscopy (EDS), flow cytometry with DNA straining, loss on ignition and dehydrogenase activity have been used by different authors to study the development of biolayer (Chaillou et al. 2011; Ochoa et al. 2015; Sabogal-Paz et al. 2020).

Many studies made use of diffusers and inlet arrangements specifically in intermittent filters to avoid disturbance to the biolayer during feeding (Dalahmeh et al. 2012; Al-Ismaili et al. 2017; Spychala et al. 2019). This aspect also needs further investigation. A few studies utilised fine and coarse media filters and multimedia filters for treatment of greywater (Babaei et al. 2019; Subramanian et al. 2020; Prabha et al. 2021). However, effect of filter depth and mode of operation on the performance of these filters need to be studies further.

Filter performance generally improves with time due to filter maturation. There is a need to develop methods to accelerate filter maturation. Methods such as feeding of wastewater and spiking of bentonite clay have been used to speed-up ripening of filters (Alsulaili et al. 2017; Babaei et al. 2019). This concept needs further study. Further, cleaning procedure of filter and deterioration of filter performance following cleaning (Katukiza et al. 2014a; Yaseen et al. 2019; Shaikh & Ahammed 2021b) also need further investigation. Studies on the influence of depth of filter on the performance are scarce (Katukiza et al. 2014a; Ochoa et al. 2015). Effect of filter depth on the removal of different pollutants including nutrients need to be studied in detail. This will help optimising the operation of filters. Different operational parameters of the filters such as HLR, OLR, media characteristics such as media size and media depth need to be optimised with respect to greywater characteristics to meet the different reuse standards.

Use of products such as soaps, detergents, preservatives, toothpaste, dish washing liquid and pharmaceuticals in the households adds a large number of organic micropollutants to the greywater. Eriksson et al. (2002) reported the presence of 900 different types of xenobiotic organic compounds (XOCs) in greywater from Danish households. Presence of different types of parabens in hand soap, showering and bathing products has been reported in multiple studies (Andersen et al. 2007; Eriksson et al. 2009; Hernández Leal et al. 2010; Hernández-Leal et al. 2011). No studies have been reported on the fate of these organic micropollutants during granular media filtration. Hence, the effects of media characteristics, loading rate and mode of operation on the removal of these pollutants should be studied in long-term filtration tests.

Presence of elevated levels of metals such as Cu and Zn has been reported in greywater in multiple studies as they are associated with leaching from galvanized tanks, fittings, plumbing materials, coatings and pipes (Christova-Boal et al. 1996; Hernández Leal et al. 2007; Al-Ismaili et al. 2017; Leong et al. 2018; Noutsopoulos et al. 2018). Therefore, it is necessary to study the potential of filters under different loading conditions to remove these.

Greywater characteristics differ from one population to another and also with the source of greywater. Greywater from the hand basin and bathroom is lightly polluted and hence, pre-treatment in the form of settling followed by saturated coarse media filtration will be able to meet the different reuse standards. An unsaturated fine media filter will be an alternative depending on the characteristics of the greywater. For the treatment of pre-treated LGW, a filter bed with a particle size in the range of 1–4 mm and depth of at least 50 cm will help achieve reuse standards. On the other hand, kitchen and laundry greywater sources are heavily polluted and hence, coagulation, constructed wetland and coarse filtration combined with saturated fine media filtration with at least 50 cm bed depth will be needed to achieve different reuse standards. Filter media with higher SSA such as GAC will have higher adsorption sites for the removal of organic content, nutrients and microbiological parameters in dark greywater sources. Multiple studies reported that greywater characteristics vary with income of the country, age, number of occupants, and geographical locations, hence, filter media to treat greywater should be selected carefully based on pilot-scale studies (Ghaitidak & Yadav 2013; Vuppaladadiyam et al. 2019; Shaikh & Ahammed 2020). Therefore, care should be taken when comparing efficiency of different filter media.

Granular filters are good at the removal of turbidity and TSS while they are moderate in removing BOD and COD with an average removal of 50–60%. The filters are good at the removal of phosphorus and performed low to moderate in removing nitrogen content in greywater. Bacterial removal during filtration can vary widely from 0.1 to 4.3 log depending upon the operating conditions. While mode of filter operation and variations in particle size, HLR and OLR do not significantly affect the turbidity removal, organic content removal in the filter is greatly impacted by these factors. Removal efficiencies of bacterial pathogens, viruses, surfactants and organic micropollutants in the filters are unknown and need to be studied in detail. Granular filters can be one of the alternatives for the on-site treatment of greywater if pre-treatment, filter media, depth of the media and mode of operation are chosen carefully with respect to the source of greywater.

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

The authors declare there is no conflict.