Characterization of dissolved organic matter at a water treatment plant with closed systems in different seasons

The 15,310 waterworks in Japan supplied water to 124.4 million people (97.9% of the population) as of fiscal 2015 (Japan Water Works Association 2017). However, only 1,381 waterworks facilities were categorized as ‘large water supply facilities’ (i.e. serving more than 5,000 people), and these units mainly used the conventional water treatment processes of coagulation–flocculation, sedimentation, filtration, disinfection, and sludge treatment. In addition to producing purified water, these processes release water treatment plant (WTP) residuals, which are waste backwash water (WBW) and clarifier sludge (CS). These residuals consist of concentrated metals (e.g. aluminum), colloidal materials, natural organic matter (NOM), and pathogens (e.g. Giardia and Cryptosporidium) (Bourgeois et al. 2005). The volume of WBW produced by the WTPs was about 2–5% of the plant's treatment capacity, and the WBW contains 50–400 mg L⁻¹ of suspended solids (USEPA 2011) and higher concentrations of total trihalomethanes (TTHM) and haloacetic acids (HAA9) (McCormick et al. 2010). High concentrations of suspended solids in raw water result in 0.5–2% solids in CS (ASCE/AWWA 1997).

Regulations in Japan prohibit the release of untreated WTP residuals to the environment. In water treatment, closed systems are those that add WTP residuals to raw water. The addition of treated WTP residuals to raw water has beneficial and detrimental effects. Gottfried et al. (2008) found that recycling 5% or 10% of the water from WBW is beneficial in terms of removing dissolved organic carbon (DOC). However, Zhou et al. (2015) found that CS releases extracellular and intracellular organics that can cause the quality of coagulated water to decline. The recycling of WBW and of recycled combined sludge (RCS) has been found to increase the turbidity, total solids, NH₃–N, chemical oxygen demand of manganese (COD_Mn), DOC, aluminum, manganese, and cadmium in plant influent; however, such recycling was found to have no significant effect on the quality of settled water or on that of filtered water (Liu et al. 2017).

The effect of the recycling process at WTP may differ from season to season because the quality of raw water differs with differences in rainfall intensity. Heavy rainfall can cause substantial changes in the levels of suspended solids, dissolved organic matter (DOM), total phosphorus, and total nitrogen in river water (Li et al. 2005). A previous study monitored seasonal variations in the concentration and distribution of natural organic matter components at WTP with open systems (So et al. 2017). The current study focuses on characterizing DOM at a WTP that uses a recycling process to utilize treated WTP residuals as raw water in different seasons (winter and summer).

Water samples were taken from a WTP in Central Japan between 2014 and 2017 in winter (November–January) and summer (July–September). The raw water source was an irrigation channel, and the raw water was collected in a water pond before being treated. The WTP has a treatment capacity of 200,000 m³ per day using coagulation–flocculation, sedimentation, filtration, disinfection, and sludge treatment in a closed system. The sludge from the sedimentation unit and the WBW were collected and treated at sludge treatment facilities (sludge concentrating tank and dehydrator), and after treatment, the water was pumped to a receiving well as recycled water and mixed with raw water. The samples in this study were collected from the water pond (raw water), the receiving well, clarified water, post-filtration–disinfection (purified) water, and recycled water (Figure 1).

Water samples were collected in pre-washed 1-L glass bottles (cleaned of organic matter by oxidization with chlorine 24 hours before use and washed with Milli-Q water). The samples were collected from the WTP and were immediately stored in a cooling box. Some parameters (e.g. temperature, pH, dissolved oxygen (DO), oxidation–reduction potential, and electrical conductivity) were measured directly onsite. After sampling, the samples were filtered using cellulose acetate membrane filters with a pore size of 0.2 μm (Toyo Roshi, Japan). The filtered samples were stored at 5 °C before chemical analyses of parameter DOC, UV₂₆₀, excitation emission matrix (EEM), and molecular weight distribution. Ultraviolet absorbance at 260 nm was analyzed using a spectrophotometer UV–Vis (UV-1600 GLP, Shimadzu Japan), and DOC was measured using a total organic carbon analyzer (TOC-Vws, Shimadzu, Japan).

The fluorescence EEM values of the samples were analyzed using a spectrofluorometer (RF-5300, Shimadzu, Japan). Excitation and emission scans were performed at 5-nm increments between 220 and 550 nm. The obtained fluorescence intensities of the samples were normalized using the quinine sulfate unit (QSU) by dividing the fluorescence intensity values of all samples by the fluorescence intensity value of 10 ppb quinine sulfate (in 0.05 M H₂SO₄ solution) at the designated wavelengths using excitation wavelength (Ex)/emission wavelength (Em) = 350/450 nm.

Source water from the river that was collected in the water pond showed different concentrations in summer and winter. Table 1 shows that some of the parameters (e.g. turbidity, DOC, and UV₂₆₀) have lower concentrations in winter than in summer, but that DO and conductivity are higher in winter than in summer. The average DOC concentration in raw water rose to 0.86 mg L⁻¹ in summer but only to 0.56 mg L⁻¹ in winter. These trends may be attributable to rainfall and increased water temperature along with microbial activity elevating organic matter in river water in summer (Li et al. 2005; Ritson et al. 2014; So et al. 2017). The fluctuations in turbidity for raw water (97% relative standard deviation, RSD) in summer may have been caused by rainfall, and the UV₂₆₀ was twice as high in summer as in winter. The highest specific ultraviolet absorbance (SUVA) (UV₂₆₀/DOC) value for raw water was 3.17 L mg⁻¹ m⁻¹ in winter and 3.41 L mg⁻¹ m⁻¹ in summer, indicating water constituents of mostly humic-like substances and high molecular weight substances (Edzwald & Tobiason 1999).

Figure 2 shows that the average DOC concentration in winter increased slightly in the receiving well after recycled water was added, and due to the high concentration of organic matter in the recycled water (1.25 mg L⁻¹), the dissolved organic concentration increased from 0.56 mg L⁻¹ (raw water) to 0.65 mg L⁻¹ (receiving well). In summer, the DOC concentration in the receiving well fell to 0.80 mg L⁻¹ after raw water was mixed with recycled water, even though the DOC concentration was 0.82 mg L⁻¹ in the recycled water and 0.86 mg L⁻¹ in the raw water. The recycled water showed a higher concentration of DOC in winter and summer: about 1.25 mg L⁻¹ and 0.82 mg L⁻¹, respectively. The purified water showed a lower concentration of DOC for both seasons: 0.42 mg L⁻¹ in winter and 0.51 mg L⁻¹ in summer. These results agree with findings reported by Liu et al. (2017), who investigated how the recycling of WBW and RCS affected the quality of the finished water. The Liu team found that the recycling procedure did not affect DOC or UV₂₅₄, although recycling did cause slight increases in NH₃–N and COD_Mn. Every summer, the WTP operator adds about 5 or 10 mg L⁻¹ of powdered activated carbon to the receiving well to reduce the DOC concentration, and this contributed to the lower concentration of DOC in the receiving well and the recycled water.

UV₂₆₀ is a parameter of UV–Vis absorption that is commonly used in Japan in addition to UV₂₅₄. Sillanpää et al. (2015) stated that any wavelength from 220 to 280 nm is appropriate for natural organic matter measurement. The percentage of reduction of UV₂₆₀ in WTP process in summer and winter was about 74% and 66%, respectively. The addition of activated carbon promotes reductions in UV₂₆₀ values in summer.

In summer and winter, the SUVA tended to decrease after coagulation–flocculation and sedimentation (i.e. in clarified water) but increased slightly after filtration (i.e. in purified water). In coagulation and flocculation process, it is easier to remove organic matter with high SUVA (Parsons et al. 2004) by bridging or sweep flocculation (Sillanpää et al. 2015). The SUVA for the purified water was 58% lower than that for the raw water in winter and 59% less than that for the raw water in summer. Meanwhile, the low SUVA (1.25 L mg⁻¹ m⁻¹ in winter and 1.03 L mg⁻¹ m⁻¹ in summer) of the recycled water indicates that the water contains mostly non-humic substances and low molecular weight compounds (Sillanpää et al. 2015).

EEM spectrofluorometry data were analyzed using peak fitting and PARAFAC. To minimize Raman scattering, all EEM data were subtracted by using a solvent blank (Milli-Q water) (Bahram et al. 2007). In peak fitting, specific peaks were used to show component fulvic acid-like substances (Peak A, Ex 260 nm; Em 380–460 nm), humic acid-like substances (Peak C, Ex 350 nm; Em 420–480 nm), tyrosine-like substances (Peak B, Ex 275 nm; Em 310 nm), and tryptophan-like substances (Peak T, Ex 275 nm; Em 340 nm) (Coble 1996). Figure 3 and Table 2 show that fulvic acid-like substances (Peak A) in raw water had maximum fluorescence intensities in winter and summer. The fulvic acid-like substances largely consist of humic substances in natural water with fluorescence intensities about 5–25 times those of the humic acid-like substances (Thurman 1985). The fluorescence intensities of humic acid-like substances (Peak C) in the raw water was about 50% from intensities of fulvic acid-like substances, and the low solubility in humic acid-like substances due to low carboxylic content makes the fluorescence intensities lower than fulvic acid-like substances.

This research found that fulvic acid-like and humic acid-like substances other than protein-like substances were removed from raw water effectively by coagulation–flocculation (type coagulant: PACl, average doses 23.7 mg L⁻¹) and sedimentation. Protein-like substances (Peak B and T) had similar fluorescence intensities in winter and summer but the fluorescence intensities of the tyrosine-like substances were about 2–4 times those of the tryptophan-like substances. The insufficient ability of conventional processes to reduce protein-like substances results in high fluorescence intensities in purified water and recycled water.

PARAFAC models were generated by the DOMFluorv1_7 toolbox (Stedmon & Bro 2008) using Matlab R2018b (Mathworks, USA). Raman and Rayleigh scattering were removed by cutting the wavelength influenced by scatter peaks and replacing missing values with a zero value to generate the best model. Only two components were identified in this research: C1 (Ex 250 nm; Em 470 nm) and C2 (Ex 225 nm; Em 430 nm). It is suggested that they are linked to humic acid-like and fulvic acid-like substances, respectively. These constituents were found at higher concentrations in summer, but the water treatment process can reduce those substances and could deliver similar quality in both seasons.

The molecular weight distribution of DOM at the WTP was measured by using HPLC-SEC with a UV detector (Figure 4). The molecular weight distribution shows that WTP can reduce biopolymers (Peak 1) and humic substances (Peak 2) effectively in winter and summer, despite the nearly doubled peaks observed for raw water in summer. Conventional water treatment can remove high molecular weight organic matter, but it is difficult to remove lower molecular weight DOM such as building blocks (Peak 3), low molecular weight acids (Peak 4), and low molecular weight neutrals (Peak 5). Figure 3 shows that the recycled water contains mainly DOM with low molecular weights, and the characteristics of DOM were similar to those of purified water.

The computed values for molecular weight characteristics (M_w, M_n, and polydispersity) are shown in Table 3 for each season. After coagulation–flocculation and sedimentation, the values of M_w and M_n are about 14% and 19% lower, respectively, than those of raw water in the winter. Moreover, the summer shows a relatively higher decrease of 23% for M_w and 16% for M_n. In winter, the recycled water has a lower value of molecular weight than the purified water but the opposite occurs in summer.

A comparison between DOM in winter and DOM in summer shows a significant difference for raw water, but no significant difference for purified water. Water treatment can reduce DOM for humic-like substances (Peaks 1 and 2 in the molecular weight distribution, and peaks A and C in EEM), but protein-like substances (Peaks 3, 4, and 5 in molecular weight distribution, and Peak B in EEM) were difficult to remove both in winter and summer. Based on the fluorescence intensity (EEM) in winter, the DOM in recycled water shows the same intensity as that in purified water for Peaks A, C, B, and T, but only Peak B and T show the same fluorenscence intensity as that in summer. The molecular weight distribution shows raw water and receiving well water dominated by high molecular weight substances, whereas the recycled water and purified water have predominantly low molecular weight substances.

There were no significant increases in the concentrations of humic-like substances in influent water of the WTP after raw water was mixed with recycled water in the study period. This may be due to the lower flow rate of the recycled water than that of the raw water. However, higher concentrations of low molecular weight and protein-like substances were found in the recycled water, which suggests a possible increase of chlorine consumption at the disinfection process in the WTP.

The main conclusions of this research are as follows:

• The values of DOC, UV₂₆₀, SUVA, fluorescence intensity and molecular weight distribution peaks are higher in summer than in winter. Despite this, water treatment can produce purified water of the similar quality in both seasons.

• The recycled water contained mostly low molecular weight and protein-like substances. This water did not significantly affect the concentrations of humic-like substances or the high molecular weight substances in influent water of WTP.

This research was supported by JSPS KAKENHI Grant Number JP17K06616.