Greywater (GW; domestic wastewater excluding toilet water) reuse for non-potable purposes, such as garden irrigation, can decrease domestic water demand and, thus, mitigate the pressure on depleted water resources while reducing household water costs [1
]. However, untreated GW contains pathogens and other pollutants and may pose environmental and health risks if used without treatment and disinfection [2
]. Chlorination and low-pressure UV irradiation are probably the most widely used disinfection methods in small, on-site GW systems [4
Chlorine is dependably effective against a wide spectrum of pathogenic microorganisms and is considered to be a cost-effective disinfectant [3
]. In addition, residual chlorine remains in the effluent after application, ensuring continued disinfection throughout the conveyance system, reducing potential regrowth [8
]. Furthermore, chlorine dosing is flexible and can be controlled by simple, low-cost devices. Nonetheless, chlorine is toxic and corrosive; thus, its storage, shipping, handling, and application must be managed responsibly.
UV irradiation prevents replication of microorganisms through a photochemical reactions that damage their nucleic acids in either DNA or RNA [10
]. The main reasons for the use of low-pressure UV (254 nm) irradiation in small on-site systems are: (1) it does not require chemical additives (making transport, storage, and dosing irrelevant), (2) it has been found to be effective on numerous pathogens including viruses and protozoans which were found to be chlorine-resistant, (3) it is cost-effective on both initial capital investment and operational levels, and (4) both the equipment operation and maintenance are simple and safe [4
It should be noted that varying disinfection efficiencies have been reported in full-scale installations [11
], and some studies have demonstrated that sub-standard water quality can reduce the efficiency of both chlorine and UV disinfection. Specifically, the presence of particulate matter and organic substances in the water may negatively impact these disinfection methods’ performances [4
]. In the case of chlorination, this negative effect is expressed by increasing the chlorine demand as dissolved and suspended organic matter is oxidized by chlorine. Thus, the overall disinfection efficiency decreases. Microorganisms attach to particles present in water, thus reducing the chance for efficient contact between the microorganism and the chlorine, compared to non-attached bacteria [12,13
]. Even more, the presence of organic matter may further reduce chlorine disinfection efficiency by stabilizing microbial cell membranes [14
]. Lastly, the presence of organic matter may lead to the formation of unwanted disinfection by-products (including known or suspected carcinogens), thereby not only impeding the disinfection process [15
] but also posing a health hazard. Winward et al. [12
] investigated the effects of organic and particulate matter on GW chlorine disinfection in a batch system, and claimed that an increase in the organic matter enhanced the chlorine demand but did not affect total coliforms’ resistance to chlorine. However, these authors recommended removing organic matter prior to chlorination in order to reduce the chlorine demand and the potential for microbial regrowth.
In the case of UV irradiation, particles interfere with the exposure of the target microorganisms to the irradiation [15
], either by shielding them, or by absorbing or scattering the light, thus reducing the UV dose received by the microorganisms and, consequently, the method’s efficiency. The presence of particulate matter and organics in GW has been noted in many studies, yet only a few discussed their adverse impact on UV disinfection efficiency. For example, the authors of [4
] studied disinfection of artificial GW, proposed limits of 60 mg/L of suspended solids and a turbidity of 125 NTU, beyond which GW cannot practically be disinfected to achieve a 4-log reduction of fecal coliforms (FC), regardless of UV reactor dimensions. Ref. [18
] recommended removing particles through filtration to obtain a turbidity level of 2 NTU (Nephelometric Turbidity Units), to increase UV disinfection efficiency. Other studies have focused on the particle sizes that block microorganisms from UV light [12
], and the specific particle types associated with certain bacteria in treated GW that cause bacterial shielding from UV disinfection [19
Interestingly, no systematic information exists regarding the combined impact of suspended solids and organic matter (measured as 5-d biochemical oxygen demand (BOD5)), on low-pressure UV disinfection and chlorine disinfection in both batch and continuous-flow disinfection units. This study aimed to test the efficiency of both disinfection methods on GW under a range of total suspended solids (TSS) and BOD5 concentrations. Experiments were performed in controlled batch laboratory setups and in flow-through reactors. Moreover, the study’s objectives included the development of regression models to predict the impact of TSS and BOD5 on chlorine and UV disinfection efficiency in both setups.
2. Materials and Methods
The research was performed in two stages. Initially, treated GW samples, varying in their TSS and BOD5 concentrations, were disinfected in a batch setup by either a hypochlorite solution or UV radiation using a collimated beam. The results obtained from this stage were used to develop two multiple linear regression models (one for chlorination and the other for UV irradiation). In the second stage, treated GW, from on-site treatment systems (outlined underneath), was disinfected in a flow-through disinfection unit using one of the two methods: chlorine tablets or commercial low-pressure UV. The models’ applicability and verification were studied, and then compared with the results of the second stage.
2.1. GW Treatment System
Eleven single-family full-scale recirculating vertical flow constructed wetland (RVFCW) systems were used for treating domestic GW (Figure S1
). The RVFCW system comprised two 500-L plastic containers (1.0 m × 1.0 m × 0.5 m) placed on top of each other. The top container that had a perforated bottom held a planted three-layered bed, while the lower container functioned as a reservoir. The bed consisted of a 10-cm lower layer of limestone pebbles, with a 35-cm middle layer of tuff gravel, and a 5-cm upper layer of woodchips. GW was pumped from a 200-L settling-equalization tank from which it was conveyed to the top of the bed. From there, it trickled through the bed layers (unsaturated flow) and into the reservoir. GW was recirculated from the reservoir to the upper bed at a rate of about 300 L/h for 8 h, after which it was filtered through a 130-μm filter and then reused for garden irrigation. Additional details about the system can be found in [20
2.2. Batch Experiment
Treated domestic GW samples (1 L) from the 11 RVFCW were collected at least four times along the study and brought to the laboratory shortly after collection in a cooler. The quality of the pre-disinfected treated GW was examined for the following parameters: TSS by the gravimetric method, BOD5
using standard 300-mL bottles, % irradiation transmission at 254 nm by a spectrophotometer (Genesys 10, Thermo), turbidity using a HACH 2100P Turbidimeter, and FC by membrane-filtration methods using mTEC agar (Lesher, Michigan USA, Acumedia). All analyses followed standard procedures [22
Treated GW samples were examined either as is or after they were subjected to concentration increases in either TSS (final TSS concentrations ranging from 1–130 mg/L) or organic matter (measured as BOD5 with concentrations ranging from 3–100 mg/L) or a combination of both suspended particles and organic matter concentrations at different ratios. Increasing the TSS was conducted by adding different amounts of powdered dried suspended solids to the treated GW. The suspended solids were prepared by concentrating raw GW (centrifugation at 6000 rpm for 5 min) and drying the pellet at 60 °C for 48 h. Organic matter concentration was increased by introduction of different quantities of 0.2 μm- filtered raw GW with known BOD5 concentrations to the treated GW. The required components were stirred in a beaker for 15 min to produce a uniform mixture. Additionally, FC were introduced by adding < 0.5 mL/L GW sample of kitchen effluent to ensure FC concentrations of 104 to 105 CFU/100 mL. Overall, 432 combinations were tested.
Subsamples were analyzed before and after disinfection, when disinfection efficiency was determined by calculating the log inactivation of FC.
2.2.1. Chlorination Experiment
The efficient application of a disinfection agent should take into consideration the required dose, which can be achieved by varying the chlorine concentration and disinfection contact time. The required dose varies based on chlorine demand (wastewater characteristics) and residual chlorine requirements. According to [23
] the free residual chlorine concentration should be ≥0.5 mg/L after at least 30 min contact time at pH < 8.0. Subsamples were disinfected in a batch mode. Initially, the chlorine demand of the subsamples was determined. For this, aliquots of 25 mL were exposed to four different chlorine doses of 0.5, 1, 3, and 6 mg/L. Samples were gently stirred and after 1 h, the total and free residual chlorine levels were determined by the DPD method [22
2.2.2. Collimated Beam Setup
A quasi-parallel beam bench-scale UV apparatus (Trojan Technologies Inc., Ontario, Canada) was used to test the efficiency of UV disinfection (Figure S2
). The system consisted of an 11-W low-pressure mercury vapor germicidal UV lamp, emitting monochromatic UV radiation at 254 nm directly over an inner 25-cm-long non-reflective collimated beam with a diameter of 40 mm. An ILT 1700 radiometer (International Light, Peabody, Massachusetts, USA) with a detector sensitive at 254 nm (IL photonic SED240) was used to measure the intensity of the incident UV light. Samples (25 mL aliquots) were placed under the collimation tube in a 50 × 35 mm crystallization dish and mixed with stirring bar (~110 rpm) allowing a uniform UV dose application to the entire sample.
Control over the UV dose was conducted by a shutter that allowed changing the exposure time of the stirred sample. Samples were exposed to three UV irradiation doses: 7.5, 15, and 30 mJ/cm2
. Exposure times for each UV dose depended on several factors, including: incident intensity, reflection, petri factors, divergence, and water factors. The methods used for determining these factors are described in [24
]. The divergence and reflection factors were constant in all experiments, and their values were 0.960 and 0.975, respectively. The petri factor was calculated every week and averaged 0.88 ± 0.05. The water factor varied from 0.40 to 0.89, and the incident intensity measured at the water surface varied from 0.30 to 0.32 mW/cm2
2.3. Flow-Through Setups
Treated GW samples (10 L) were taken from the on-site full-scale single-family RVFCW (Section 2.1
above), immediately transported to the laboratory, and served as inflow to the continuous-flow disinfection units. All samples were analyzed for TSS, BOD5
, % irradiation transmission at 254 nm, turbidity, and FC as described above. After disinfection, samples were analyzed again for FC.
2.3.1. Flow-Through Chlorination Chamber
Chlorination was performed by discharging the treated GW (at a predetermined flow rate) via a 500-mL chamber containing a slow release HTH tablet (High Test Hypochlorite; 70% available chlorine, HYDRO-LINE, Silinierby, Finland). The chamber was a 500-mL Amiad filter housing without the filter (Model. BSP 1″, Amiad Ltd., Amiad, Isael; Figure 1
a). A single chlorine tablet was placed in the flow-through chamber and was designed to dissolve slowly as the water flows through the chamber, according to the determined contact time. The chamber was connected at both ends to tubes; the input tube was connected to a submerged aquarium pump (Atman, model AT 102, Guangdong, China) that regulated the input flow at 8 L/min, mimicking typical rates in regular GW reuse garden systems. In other words, each sample of treated GW was exposed to the same contact time, although the quality of the treated GW was quite different, and, thus, there could be large variability in the required chlorine dose. Chlorinated samples were collected from the outlet tube.
2.3.2. Flow-Through UV Reactor
A low-pressure continuous flow UV reactor (UV6A, WaterTec Inc., Pan- Chiao Taipei, Taiwan) with a startup time from turn-on to maximum intensity of 100 s was used to irradiate samples (Figure 1
b). The reactor (43 mL in volume) contained a low-pressure 4 W mercury lamp and was 1.6 cm in diameter and 13.5 cm in length. More details about the UV reactor can be found in [25
]. The lamp was turned on at least 120 s after which treated GW samples were pumped through the reactor using a peristaltic pump (Masterflex, Cole-Parmer Instrument Co., Chicago, IL, USA) at a flow rate of 24 L/h. The iodide–iodate chemical actinometry (for details see [25
]) was used to determine the actual average UV dose in the reactor which was found to be 44 mJ/cm2
, with a calculated lamp intensity of 2.8 mW/cm2
and a mean residence time of 14 s.
2.4. Multiple Linear Regression (MLR) Models
The results of the batch experiments were used to develop MLR models. The models are intended to predict the log inactivation of FC based on water quality parameters and applied disinfectant dose (chlorine or UV irradiation). The water quality parameters chosen for the model (TSS, BOD5 and log FC concentrations in the GW prior to disinfection) were expected to significantly affect the model prediction and the coefficient of determination (R2). The developed models were validated against the results from the on-site greywater samples that were obtained from the flow-through reactor experiments. Finally, the models were used to propose a conversion factor between the chlorination or UV collimated beam batch laboratory setups and the experimental continuous-flow reactors results.
This study quantified the effects of treated greywater quality (TSS, BOD5, and FC) on both chlorination and UV disinfection efficiencies in batch and continuous flow setups.
The efficiency of chlorine disinfection of treated GW was found to decrease as a result of increasing TSS and BOD5 concentrations, in which the effect of TSS was continuous starting from low concentrations, while the effect of BOD5 became significant only above a certain threshold concentration. Batch chlorination experiments have shown that dissolved organic matter affects chlorination efficiency significantly less than TSS, as reflected by the much lower LogWorth value. Based on the batch chlorination results, an MLR model was developed and successfully verified against the results of a flow-through chlorination unit.
The results of the batch UV disinfection experiments suggest that the UV disinfection efficiency of treated GW decreases as a result of increasing TSS concentrations beyond a threshold value of 50 mg/L. Yet, as the applied UV dose increased, the influence of TSS decreased. The effect of dissolved BOD5 on UV disinfection efficiency was found to be negligible (in the concentration range tested).
Similarly, based on the batch UV disinfection experiments an MLR model was developed and was verified against the results of treated GW that was disinfected by a flow-through UV reactor. Using these two models, one can assess the UV dose, or the residual chlorine concentration required in flow-through reactors based on batch results. This approach is valuable not only from an operational standpoint, but also from a research perspective.