Effects of adsorption and filtration processes on greywater microbiological contamination and the potential human health risk reduction

Water scarcity is a significant issue impacting economic development, environmental sustainability, and quality of life. The sustainable usage of existing water resources is key to addressing the challenge of rising water demand. Black water originating from water closets contains significant amounts of pollutants in comparison to greywater (GW), which is defined as urban wastewater that excludes any contribution from water closets and includes wastewater (WW) from baths, showers, hand bowls, laundry, dishwashers, and kitchen sinks (Eriksson et al. 2002). Some literature (Al-Jayyousi 2004) excludes the contribution of kitchen WW from other GW streams. Given that approximately 50–80% of the WW is categorized as GW, one of the many strategies for water management can be recycling of GW for non-potable purposes. Onsite GW reuse after treatment can potentially mitigate water shortage issues.

Although GW reuse is critical, there are concerns about reclaimed water quality, human health perceptions, and economic considerations. The GW quality can vary with the source, geographical location, lifestyles, traditions, installations, level of occupancy, demographics, and chemical usage (Eriksson et al. 2002; Al-Jayyousi 2004). Additionally, biological and chemical degradation during storage and transportation can further alter the GW quality. In GW, total coliforms (TC) range from 101 to 107 CFU/100 mL (Eriksson et al. 2002). The concentration of coliforms significantly depends on the demographic distribution of the individuals. High concentrations of total and fecal coliforms have been estimated in the GW generated from a household comprising of young children with values of 3.2 × 10⁵ and 1.5 × 10³ CFU/100 mL, respectively, in comparison to 80 and 6 CFU/100 mL for a household without children (Rose et al. 1991). Commonly, microbiological contaminations are measured using Escherichia coli, TC, and fecal coliform (FC) indicators. E. coli is considered as the most common pathogenic indicator of pollution (Odonkor & Ampofo 2013). Also, the presence of E. coli is often associated with the potential presence of other pathogenic organisms, including Salmonella spp. or hepatitis A virus (Brüssow et al. 2004; Truchado et al. 2016). These bacteria are induced in the GW from washing hands after using latrine, washing infant's contaminated clothes including diapers, and washing food waste from the plates and utensils. The presence of these different microbiological contaminants makes the raw GW reuse undesirable. This often remains a key barrier to a wide-scale implementation of GW reuse. The treatment of GW has the potential to address the concerns of microbiological and other concerns of GW reuse.

GW treatment can be carried out onsite or at a separate, centralized location within a community. However, the treatment system should be sustainable, and the reclaimed water quality should meet the standard reuse guideline for health and safety. Among numerous available water treatment technologies, adsorption using solid materials is considered to be the simplest and most effective process (Crini et al. 2019). Activated carbon (AC) is an extensively and widely used adsorbent material for industrial-scale water treatment. However, although AC is a preferred adsorbent, its widespread application is limited due to it being quite expensive (Crini et al. 2019). Alternative materials, like ground blast furnace slag (GBFS), can act as a viable option (Zannerni et al. 2020). Even though the utilization of GBFS for the extraction of heavy metals and phosphorus from aqueous solution and wastewater has been documented (Mortula et al. 2007), no studies have evaluated the use of GBFS in the treatment of microorganisms. Apart from adsorption, there are other technologies used previously for GW treatment. An experimental study was carried out to assess the use of coagulation and flocculation on shower GW in the United Kingdom (Pidou et al. 2008). Both ferric sulfate and aluminum sulfate displayed comparable TC and E. coli removal efficiencies of more than 99.9%. A study by Sinno et al. (2022) found that electrocoagulation was capable of substantially removing turbidity, total organic carbon, chemical oxygen demand and grease. One of the studies evaluated the recycled vertical flow bioreactor by treating synthetic GW (Gross et al. 2007). The experimental results demonstrated that the E. coli reduced from 501,188 to 1.25 CFU/dL. Overall, the removal efficiency of filtration and membrane bioreactor treatment systems is 100% for E. coli and 99.9% for TC (Ghaitidak & Yadav 2013).

GW reuse is always associated with concerns of potential contamination and microbial regrowth and eventual spread of diseases. Different researchers have investigated and assessed the impact of GW reuse on human health. Untreated GW reuse often poses an unacceptable risk to human health from wide-ranging levels of exposure from disposal to reuse. Nunez et al. (2014) evaluated the risk of human health in Argentina due to untreated GW disposal practices. The study observed the presence of indicator organisms in the disposal sites and identified that the filtration process removes the microorganisms in significant numbers. Neto et al. (2018) investigated the risk of treated GW usage and observed toilet flushing exposure to be the highest microbial risk. Kusumawardhana et al. (2021) used a quantitative microbial risk assessment (QMRA) approach for conducting mathematical health risk estimation. The study investigated both rainwater and untreated GW for the risk posed by Staphylococcus aureus, E. coli, Salmonella spp., and Giardia lamblia, among others. They reported the necessity of appropriate GW treatment for acceptable risk. Shi et al. (2018) also used the QMRA approach for the risk estimation of treated GW. The study focused on the exposure of treated GW pollutants during toilet flushing and food-crop irrigation. Treatment using microfiltration was capable of ensuring an acceptable risk level for toilet flushing. Blanky et al. (2017) used the QMRA approach to evaluate the risk for toilet flushing and gardening from Legionella only. The researchers did not find significant risk reduction from the use of treated (using constructed wetland) and chlorinated GW for garden irrigation and toilet flushing.

Despite the obvious need, there have been limited studies on the treatment of GW for reuse purposes. Also, even though some of the individual treatment processes were investigated in terms of their efficiency of pollutant removal, the adsorption process was not examined for removing microbial contamination and the eventual risk. The primary goal of this paper is to investigate the effectiveness of microbial inactivation and risk reduction using AC, GBFS and sand filtration (SF) in treating GW. Pilot-scale experimental setup was used to treat GW generated in a large apartment complex. The QMRA approach was used to conduct the risk estimation for FC and E. coli.

For this study, a 30-floor residential tower (Al Jawad Building) located in the Al Nahda area in Sharjah was chosen. The building was chosen as it already has an operational large-scale onsite GW treatment facility, which treats residential GW generated from baths and toilet sinks. The system proposed in this study was set up from the existing treatment system using a combination of different bypass pipes.

To meet the objective of this study, three materials (AC, GBFS, and filter media (silica sand)) were used. AC and silica sand were used from Jacobi Carbons and National Factory for Processing and Treating Minerals, respectively. Table 1 shows the properties of AC, GBFS and silica sand.

While the commode flush water (black water) directly discharges into the sewer system, the water generated from the bathroom sink and bath discharges into a separate pumping system installed only for the collection of GW from these two sources. During the days when the volume of generated GW is higher than the treatment system's capacity, the excess GW overflows into the sewer system without treatment.

The treatment process involved the collection of GW in an aerated underground storage tank. Two pumps (one duty and one standby) were used to lift the water into different treatment process designs. The filtration and adsorption columns had sizes of 137 mm in height and 33 mm in diameter and were operated with a maximum flow of 1 m³/h at a pressure of 2.5–3.0 bar. Sodium hypochlorite was used for disinfection to maintain a residual chlorine level of 1 mg/L in all the setups. All the treatment designs were operated at a constant hydraulic loading rate (HLR) of 11.7 m³/m²/h (or 280.8 m³/m²/day). The HLR was chosen in line with the existing GW treatment system within the apartment complex. Filtration and adsorption systems were back washed every day for 10 min at a flow rate of 2.5 m³/h using effluent from the media.

The third design relied on AC adsorbent for the treatment of the GW without any pretreatment using filtration (Figure 1(c)). Hence, the GW influent is directly fed into the AC column. Treatment also used a UV disinfection before the chlorination process. UV sterilizer with a treatment dose of 40 mJ/cm² was used in the treatment designs. The process design is similar to the first design except for the use of a filtration system, including alum pretreatment. The fourth design was similar to the previous (design 3), except GBFS was used in place of AC columns (Figure 1(d)). This was done mainly to determine the microbial removal efficiency of GBFS without any dependency on pretreatment.

Grab samples were collected every day at 6 pm (±1 h). Duplicate samples were collected from each sampling location to ensure the reliability and accuracy of the result. Sampling bottles for microbial analysis were sterile, to ensure there was no bacterial contamination, and sodium thiosulfate was used to neutralize the effect of any chlorine at the time of collection. For this reason, 50 mL sterile bottles with sodium thiosulfate were used. The sampling points were cleaned with 70% ethanol to ensure minimal bacterial contamination. All the samples for microbial analysis were tested on the day of collection.

Three types of microbial analysis including TC, FC, and E. coli were conducted in this study. Standard analytical techniques were adopted for analysis (APHA 2012). The presence of TC and FC is detected using the ‘most probable number’. On the other hand, the presence of E. coli was detected using the ‘spread plate method’. Table 2 shows the culture medium, technique, standard, and colony color used for assessing different microorganisms. Samples from each sampling location were collected three times a week and tested for microbial contamination.

The TC, FC, and E. coli analysis required the preparation of Petri dishes with a culture medium. For TC and FC culture media, a total of 3 L of culture medium was prepared. Each liter of culture medium was prepared by adding 33.12 g of TTC Tergitol-7 agar base in 1,000 mL of distilled water and heated to the boiling point to homogenously dissolve the medium. For E. coli analysis, a total of 2 L of culture medium (TBX agar) was prepared by suspending 36.6 g of TBX agar in each liter of distilled water. The prepared agar solution was autoclaved at 15 lbs pressure (121 °C) in a sterilized bottle and cooled to 45–50 °C. The solution (15 mL) was then poured into petri dishes and covered the circumference of the dish.

Since the GW samples were highly contaminated with bacteria, dilution of the samples was necessary. An ideal result is achieved when the number of colonies lies within the range of 20–80 colonies per membrane; however, the number of colonies should not be more than 200 (APHA 2012). The dilution factor varied (from no dilution to 1,010) at different sampling points depending on the microbial contamination of the sample.

Microbial enumerations are critical for risk estimation. Even though E. coli is the most specific type of organism used for indicating pathogenic organisms, many wastewater treatment plants around the world only monitor TC and FC. In this study, the QMRA model was used for both FC and E. coli, as they are known as indicator organisms.

Table 5 presents mean influent TC for all the treatment process designs. The TC ranged from 1.0 × 10¹² to 1.8 × 10¹⁴ CFU/100 mL. The values are higher than those in previous studies (1.2 × 10³–8.2 × 10⁸ CFU/100 mL) (Masi et al. 2010). The high values may be due to, but not limited to, hand washing activity after toilet use, washing of babies or diapers of babies in the bath or sink, age distribution and number of household members, traces of urine, dead skin, sweating from body, and usage pattern and season (Blanky et al. 2015).

Results demonstrated that the GBFS achieved significantly higher log reductions for TC than AC (Figure 2). GBFS reduced TC in the range of 0.5–2.1 log CFU/100 mL, compared to 0.2–0.5 log CFU/100 mL for AC. GBFS demonstrated higher TC removal of 97.8, 99.1, and 71.3% on days 1, 2, and 3, respectively, compared to 40.0, 40.0, and 66.7% on days 1, 2, and 3, respectively, for AC. This is consistent with the findings from TC removal with the filtration process (designs 1 and 2). The results are also consistent with the previous literature, where AC is capable of limited TC removal (Hijnen et al. 2010). The authors identified that the removal efficiencies are high for native coliform species. The presence of surface charge did not seem to have a significant impact on the TC reduction efficiency. The low (up to 0.5) log reductions from AC can be attributed to the fact that only the physical straining process has been effective and other surface-based processes did not contribute much to the TC removal mechanism. TC is typically made up of wide-ranging microorganisms, and AC was not capable of removing diverse groups of microorganisms well. Regarding the GBFS, the results are consistent with the literature, where 2 log reductions of TC were achieved using blast furnace slag (Abdolahnejad et al. 2017). However, the study mostly operated in slow medium filtration as opposed to rapid filtration in this study. For this reason, the result in this study is not likely to be due to schmutz deck media, which is typical for slow sand filters. It is possible that the presence of heavy metals in the blast furnace slag inactivated wide-ranging microorganisms. While it was not part of the current study, future investigations are planned to determine the impact of GBFS composition on the inactivation of different microorganisms.

The study revealed the role of filtration in enhancing TC reduction. Figure 2(a) compares the TC log reduction achieved by both SF + AC and AC alone, whereas Figure 2(b) exhibits the TC log reduction calculated for SF + GBFS and GBFS alone. While the TC reduction ranged between 0.2 and 0.5 log CFU/100 mL with AC alone design, this reduction was improved to 0.4–1.7 log CFU/100 mL with the inclusion of a filtration system. A similar pattern was noticed with GBFS too. The TC removal with GBFS varied between 0.5 and 2.0 log CFU/100 mL, eventually improving to 3.9–4.8 log CFU/100 mL with the addition of the filtration system. The results (0.1–3.4 log CFU/100 mL) are consistent with the previous studies that ranged between 0.6 and 4 log CFU/100 mL (Li et al. 2012). Although the previous literature by Li et al. (2012) identified the processes to be improved straining process with time, quick ripening, and heterogeneity in the media (Williams et al. 2007), however, the reduced TC removal efficiencies toward the end of 7 days were noticeable in this study. The reduction in filter efficiency is a common phenomenon as the filter media becomes coated with flocs, plugging the voids between the filter grains, thereby reducing the straining capacity and adsorption ability of the media. Also, during the experiments on the 7th day, it was far from the last backwash conducted on the media. This indicated that the filter media were saturated.

Table 5 presents mean FC concentrations for GW influents for all the different designs. The FC varied in the range of 7.0–8.9 log CFU/100 mL. This is consistent and a bit higher than the previous studies that determined up to 6 log CFU/100 mL (Masi et al. 2010)_. This high value could be due to the presence of a nursery facility in the building, increasing the possibility of fecal contamination due to the washing of babies or diapers in the bath or sink.

Figure 3 illustrates the FC removal efficacy achieved by both AC and GBFS. Results demonstrated that AC without any filtration achieved a reduction of 2.2, 2.2 and 1.9 log CFU/100 mL for days 1, 4, and 7, respectively. For a similar design, GBFS successfully removed 2.3, 2.2 and 2.0 log CFU/100 mL on days 1, 4, and 7, respectively. The performance of the treatment system used in this study was better than a previous study on GW treatment using an aeration process followed by AC adsorption, which achieved the FC removal of approximately 45–56.7% (Radhi & Borghei 2017). GBFS was capable of removing FC better than AC. It is consistent with the findings from FC removal with the filtration process (designs 1 and 2). As reported earlier, AC has limited ability to perform on coliform organisms. However, the FC log reductions for AC were better than TC log reductions. It is consistent with the notion that the physical straining process is more suitable for FC than other coliform organisms. The FC log reductions of GBFS in both the designs (with and without filtration) were also higher than those for TC log reductions. The high FC log reductions for GBFS compared to AC may be due to the presence of heavy metals and metal oxides in the blast furnace slag that can cause FC to inactivate and eventually be removed. The adsorptive capacity of GBFS in fixed bed column experiments is better than that of AC, indicating the reason for the improved FC reduction (Mortula & Gagnon 2007).

Similar to TC, results demonstrated that filtration significantly improved the FC removal efficiency of the system. While the reduction ranged between 1.9 and 2.2 log CFU/100 mL with the AC alone, this reduction range was improved to 3.3–4.6 log CFU/100 mL with filtration and AC. Similarly, the use of GBFS alone achieved FC removal efficiencies of 2.0–2.3 log CFU/100 mL, which was improved to 5.3–5.9 log CFU/100 mL for the system with filtration and GBFS. Hence, the overall removal efficiency of the filtration process can be attributed to 1.3–2.5 and 3.3–3.6 log CFU/100 mL when combined with AC and GBFS, respectively. A similar reduction in FC removal efficiency (62.5–100%) has also been reported in several studies conducted on SF (Radhi & Borghei 2017; Sharath et al. 2017). Since the process used in this study was rapid SF, biological layers are highly unlikely to remove microorganisms, commonly observed in slow sand filters. FC log reductions using filtration are higher than TC log reductions. It is consistent with other adsorbents in that filtration can remove FC better than other coliform organisms. Unlike TC log reductions, there was no consistent decrease in FC removal on the 7th day. As was the case for decreased efficiency due to clogging, the impact was not that considerable for FC removal performance.

Table 5 also provides the mean E. coli numbers as CFU/100 mL quantified for GW influent corresponding to all the designs. The influent E. coli in this study ranged 3.7–4.2 log CFU/100 mL, and these are consistent with previous literature on GW (Atanasova et al. 2017).

There were no E. coli bacteria observed in the effluents of both the treatment processes (filtration and adsorption) investigated in this study. It indicates that the efficiencies of E. coli removal using filtration combined with AC and GBFS were observed to be at 100% (Table 6). However, due to zero E. coli numbers in the effluents, the study could not examine the efficiencies of the filtration process alone. There were no studies examining both the filtration and adsorption processes together. Several studies have reported achieving 100% E. coli removal efficacy using filtration (Ghaitidak & Yadav 2013). However, some other studies reported that the E. coli removal efficiency of filtration ranged from 95.0 to 98.5% (Sharath et al. 2017). As observed, the influent concentrations were in the range of 10³–10⁴. For this reason, 3–5 log reductions of E. coli would be sufficient to remove the microorganism completely. It indicates that prior treatment using adsorption and/or filtration was capable of reducing E. coli completely. E. coli log reductions are similar to the FC log reductions and higher than the TC log reductions for the same experiments. It can be indicative of effective removal of E. coli using fixed bed column experiments and FC removal is predominantly due to E. coli removal. As there was not much characterization carried out in this study, it is difficult to make a conclusive statement. However, it could be due to the high proportions of E. coli in the FC.

The effluents of AC and GBFS (designs 3 and 4) treatments had zero E. coli (Table 6). It is noticeable and significant that despite variations in the influent E. coli numbers, all the setups achieved 100% E. coli removal. The adsorbents alone displayed excellent E. coli removal efficiency without prefiltration. Previous studies indicated that the presence of titanium in blast furnace slag was capable of inhibiting E. coli growth (Westholm et al. 2010; Ang et al. 2013). Westholm et al. (2010) observed 6 log reductions (higher than the present study) of E. coli to achieve complete removal. However, the process of E. coli removal using GBFS is not well explored. However, the high percentage of heavy metal in GBFS may have enough toxicity against E. coli. E. coli removal using AC was not that significant in another study (Hijnen et al. 2010). However, they could not identify AC providing significant barriers during water treatment. However, the study was focused on drinking water treatment, not identical or similar to the GW treatment in this study. Another study observed 2 log reductions of E. coli using unmodified AC (Pal et al. 2006). However, the differences in results in this study can be attributed to the type of AC. Depending upon the formation, there can be many different types of AC. So, there can always be some discrepancy in comparing the results.

The study examined the effect of filtration and adsorption processes for treating microorganisms from GW using large-scale experiments. The efficacy of SF combined with AC and GBFS in the reduction of TC, FC, and E. coli was evaluated. The study used the QMRA approach to evaluate the risk (beta-Poisson and exponential models) reduction due to GW treatment using adsorption and/or filtration. The TC removal achieved by SF + AC and SF + GBFS was in the range of 0.4–1.7 and 3.9–4.8 log CFU/100 mL, respectively. Likewise, SF + GBFS showed higher FC removal efficacy (5.3–5.9 log CFU/100 mL) than SF + AC (4.1–4.3 log CFU/100 mL). The filtration process was also capable of removing TC substantially. Adsorption treatment without a prior filtration process demonstrated high log reductions, however, lower than that with the filtration process. The AC and GBFS were very efficient in the reduction of TC. GBFS displayed higher percentage removal efficiency of 97.8, 99.1, and 71.3% on days 1, 4, and 7, respectively, than AC treatment of 40.0, 40.0, and 66.7% for the same days. AC achieved a reduction of FC by 2.17, 2.18, and 1.94 log CFU/100 mL for days 1, 4, and 7, respectively. GBFS successfully removed 2.30, 2.20, and 1.96 log CFU/100 mL on days 1, 4, and 7, respectively. The filtration process was capable of achieving substantial FC log reductions. GBFS also demonstrated higher FC log reductions than TC log reductions. FC log reductions were higher in the treatment designs than TC log reductions. It is indicated that fixed bed column filtration systems were more effective in removing FC than other coliform organisms. All the treatment designs were capable of completely removing all the E. coli bacteria. No E. coli strains were detected for any of the configurations after either adsorption alone or with filtration. The study also concludes that GBFS (with or without the inclusion of a sand filter) achieved higher microbial removal than AC. Lastly, the risk reduction achieved by AC and GBFS through exposure to FC and E. coli was substantial and often reaches acceptable limits. The P_inf.annual results demonstrated that GBFS alone had produced effluent quality with lower health risk in comparison to AC. However, the addition of filtration was capable of improving risk reduction substantially. As the process design suggests, the use of chlorination/UV disinfection was capable of removing the microbial contamination completely, virtually eliminating the risk of E. coli contamination from toilet flushing.

The authors would like to acknowledge the Enhanced Faculty Research Grant (EFRG2-SRC-CEN-32 and FRG20-M-E44) and Open Access Program (OAPCEN-1410-E00177) in the American University of Sharjah for supporting the study. This paper represents the opinions of the authors and does not mean to represent the position or opinions of the American University of Sharjah.

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

The authors declare there is no conflict.