Feasibility Study on Combined Recirculation of Production Wastewater at Water Treatment Plants

ABSTRACT

This study investigates the feasibility of reusing production wastewater from drinking water treatment plants through combined recirculation of sedimentation sludge water (SSW) and filter backwash water (FBW). Jar tests and response surface methodology (RSM) were employed to optimize the recirculation point, coagulant dosage, and recirculation ratio. Single‐factor experiments revealed that direct recirculation of SSW and FBW at 4%–6% achieved optimal turbidity removal, while recirculation of their supernatants (SSS and FBS) at 4% was more effective for UV₂₅₄ removal. RSM optimization indicated that the recirculation ratio was the dominant factor influencing pollutant removal. Under optimal conditions (predosage recirculation, PACl dosage of 16.84 mg/L, recirculation ratio of 5.38%), turbidity and UV₂₅₄ removal rates reached 94.69% and 63.87%, respectively. Notably, combined recirculation of SSS and FBS further enhanced UV₂₅₄ removal to 70.05%. The combined recirculation process also reduced coagulant consumption by 15%–20% while maintaining effluent quality. Under laboratory conditions, this study demonstrates the technical feasibility and potential energy‐saving benefits of the proposed strategy for wastewater reuse, pending pilot‐scale validation.

Summary

• Supernatant recirculation (SSS and FBS) outperforms raw sludge recirculation (SSW and FBW) in UV2₅₄ removal during drinking water treatment wastewater reuse.

• Recirculation ratio is the key process parameter governing coagulation efficiency in combined recirculation, exceeding the impact of coagulant dosage.

• Combined recirculation at 5.5% ratio with 15.8 mg/L PACl achieves > 96% turbidity removal and saves 15%–20% coagulant consumption.

Water samples were collected from the sedimentation sludge and filter backwash water of the water supply plant. A combined reflux experiment was conducted using a program‐controlled six‐stirrer system. The experimental data were then analyzed with Design Expert 13 to identify the optimal conditions.

1. Introduction

The distribution of water resources in China is characterized by considerable spatial and temporal variability, with per capita water availability remaining relatively low (Xu et al. 2015). In the process of drinking water treatment, water treatment plants generate production wastewater, which typically accounts for approximately 3%–8% of the total water supply. This wastewater consists primarily of sedimentation sludge water (SSW) and filter backwash water (FBW) (Xu et al. 2011). In addition to containing natural organic matter, this residual stream also carries substantial quantities of unstable particles, polymer colloids, and unreacted coagulants (He 2009; Sun et al. 2012).

Previous studies have demonstrated that proportionally blending production wastewater with raw water for recirculation can significantly enhance treatment efficiency (Xu et al. 2015). This improvement is attributed to the combined effects of polyaluminum chloride, hydroxide adsorption, and hydraulic flushing, which collectively improve the removal of turbidity, color, organic matter, and metallic elements from raw water. In addition to enhancing water quality, this approach has been shown to reduce residual flow treatment costs and minimize coagulant consumption.

Furthermore, the recirculation of wastewater during coagulation has been observed to promote the formation of irregularly shaped, larger flocs (Zhu et al. 2017; Qi et al. 2011; Zhou et al. 2012; Guan et al. 2005; Liu et al. 2016). Similarly, the recirculation of FBW has been associated with improved removal of dissolved organic matter, particularly humic substances (Zhou et al. 2015).

It is important to note that existing research has primarily focused on the individual recirculation of either SSW or FBW, with limited investigation into the synergistic effects and system‐level optimization achievable through their combined recirculation. As summarized in Table 1, SSW and FBW exhibit notable differences in composition and physicochemical properties. SSW is characterized by substantially higher turbidity and organic matter content, whereas FBW displays comparatively lower levels of these constituents. Such disparities may give rise to distinct yet potentially complementary mechanisms for enhancing coagulation performance during recirculation. Consequently, simply adopting optimal conditions established for single‐stream recirculation may not fully exploit the synergistic potential of combined recirculation, nor accurately identify its globally optimal operating conditions under multifactorial interactions (Kang et al. 2022; Xu et al. 2016; Daud et al. 2018; Moradi and Ghanbari 2014).

TABLE 1.

Water quality status of SW, SSW, FBW, SSS, and FBS.

Water samples		Turbidity (NTU)	pH	NH₃‐N (mg/L)	CODMn (mg/L)	UV2₅₄ (cm−1)	Al (mg/L)	Fe (mg/L)	Mn (mg/L)
SW		8.9~18.7	6.59~7.45	0.12~0.25	1.90~2.70	0.024~0.033	0.012~0.018	0.018~0.023	ND
WTRs	SSW	942.1~2960.0	6.58~7.18	0.89~1.43	6.56~7.92	0.283~0.321	0.036~0.041	0.035~0.038	0.188~0.258
	SSS	31.3~32.1	6.98~7.02	0.68~0.75	5.61~6.08	0.160~0.171	0.024~0.027	0.009~0.013	0.003~0.011
	FBW	66.0~97.3	6.89~7.32	0.41~0.48	5.21~5.38	0.189~0.220	0.023~0.029	0.007~0.009	0.004~0.012
	FBS	15.6~26.4	7.06~7.12	0.36~0.45	1.80~1.92	0.056~0.065	0.010~0.013	ND	ND

Response surface methodology (RSM) has been widely recognized as an effective tool for multifactor optimization and has been successfully applied to parameter optimization in water treatment coagulation processes (Ji et al. 2024; Birhanu and Leta 2021). The present study proposes the use of RSM to systematically investigate the interactive effects of three key operational parameters—recirculation point, coagulant dosage, and recirculation ratio—on the effluent quality of the combined recirculation process (Ji et al. 2024; Birhanu and Leta 2021; Chen et al. 2018; Kang et al. 2022).

This study offers several novel contributions: (i) a systematic comparison of the coagulation enhancement mechanisms associated with raw waste streams (SSW and FBW) and their respective supernatants (SSS and FBS); (ii) the application of RSM to holistically optimize the interacting parameters governing combined recirculation performance; and (iii) a quantitative assessment of the trade‐offs between turbidity and UV₂₅₄ removal, enabling the identification of globally optimal, energy‐efficient operating conditions.

The specific objectives of this study are threefold: first, to elucidate the effects of separate recirculation of SSW, FBW, and their supernatants on key water quality indicators and to identify the corresponding optimal recirculation ratios; second, to establish an RSM‐based optimization framework for combined SSW and FBW recirculation, characterize multifactor interactions, and determine the optimal operating conditions; and third, to evaluate the practical feasibility of the combined recirculation process in reducing coagulant dosage and achieving energy savings while maintaining effluent quality. Ultimately, this study aims to provide a system‐optimized technical solution and theoretical foundation for the resource recovery and reuse of production wastewater from drinking water treatment plants.

2. Materials and Methods

2.1. Water Sample Collection and Pretreatment

All experimental samples were collected from a drinking water treatment plant in Nanchang City, Jiangxi Province, employing conventional processes (coagulation–sedimentation–filtration–disinfection). The plant is supplied with water by the Gan River, and has a design capacity of 10 × 10⁵ m³/day. The process employs polyaluminium chloride (PAC) as the coagulant and sodium hypochlorite (NaClO) as the disinfectant. Source water (SW) was collected from the plant's raw water intake. Samples were collected on six independent sampling campaigns between March and May 2025 (spring season) to capture the variability of raw water quality. Water treatment residues (WTRs) were also obtained from the same plant, comprising SSW and FBW. SSW and FBW were placed in glass measuring cylinders. Following a period of 30 min during which the samples were left to settle, the samples were then taken 2 cm below the liquid surface in order to obtain sedimentation sludge supernatant (SSS) and filter backwash supernatant (FBS). All samples were stored in a refrigerator at 4°C until analysis, and all analyses were performed within 48 h of collection to ensure sample integrity. For each experiment, fresh grab samples of SW, SSW, and FBW were collected from the plant on the same day to ensure representativeness. All jar tests were performed in triplicate (n = 3). The interannual variation ranges of water quality parameters for SW, SSW, FBW, SSS, and FBS are presented in Table 1.

2.2. Experimental Methods and Procedures

The present study investigates the impact of varying recirculation conditions (SSW, SSS, FBW, and FBS) on the quality of effluent sedimentation. In order to do this, a series of laboratory trials were conducted using both single and combined recirculation systems. Water samples were collected from the following three distinct points: the raw water intake point, the flocculation tank inlet, and the sedimentation tank inlet. By adjusting the raw water collection location, different recirculation points were simulated: prechemical dosing, postchemical dosing but premixing, and postchemical mixing. SSW, SSS, FBW, and FBS were added to corresponding SW samples at varying ratios. The recirculation ratio was defined as the proportion of recirculated water volume to total SW volume, with six gradient settings (0%, 2%, 4%, 6%, 8%, and 10%) to cover the operational range achievable in practical water treatment plants. In order to simulate the coagulation‐sedimentation process in water treatment plants, standard beaker tests were conducted under constant temperature conditions of 25°C using a six‐station stirrer (TA6‐1, HengLing, China). A commercial grade poly‐aluminum chloride (PACl, Yu'ang Environmental Technology Co. Ltd., China) with an Al₂O₃ content of 10% was used as the coagulant. The process was comprised of five distinct stages: Stage 1 involved the mixing of SW with return wastewater (e.g., SSW and FBW) at 200 rpm for 10 min to ensure thorough blending. Stage 2 entailed the uniform dosing of PACl solution (15 mg/L concentration) over 15 s to initiate flocculation. Stages 3 and 4 constituted the coagulation‐flocculation mixing phase, divided into two steps: The experimental procedure involved initial agitation at 150 rpm for 180 s, followed by sustained agitation at 80 rpm for 1020 s. The fifth stage entailed cessation of agitation, allowing for sedimentation for a duration of 30 min, and the collection of water samples from a depth of 2 cm below the surface. Following the collection of water samples, the turbidity, ammonia nitrogen (NH₃‐N), permanganate index (CODMn), ultraviolet absorbance at 254 nm (UV₂₅₄), and aluminum (Al), iron (Fe), and manganese (Mn) concentrations in the settled water sample must be measured.

2.3. Analytical Methods

Turbidity was measured using a portable turbidimeter (model 2100P, manufactured by HACH, USA). Ammonia nitrogen (NH₃‐N) concentration was determined via Nessler's reagent spectrophotometry, while the permanganate index (CODMn) was analyzed using acid potassium permanganate titration. UV absorption at 254 nm (UV₂₅₄) of the samples was measured using a UV–visible spectrophotometer (DR6000, HACH, USA) with a 1‐cm quartz cell. The results are reported as UV₂₅₄ absorbance per centimeter (cm⁻¹). Concentrations of aluminum (Al), iron (Fe), and manganese (Mn) were determined by inductively coupled plasma optical emission spectrometry (iCAP 7000, Thermo Fisher Scientific, USA).

2.4. Response Surface Methodology for Optimization Design

Based on the Box–Behnken design (BBD) principle, experiments were conducted to investigate the combined recirculation of production wastewater, building upon the findings from single‐stream recirculation trials. The independent variables selected for the combined recirculation experiments were recirculation point (X₁), coagulant dosage (X₂), and recirculation ratio (X₃), while the response variables were turbidity removal rate (Y₁) and UV₂₅₄ removal rate (Y₂). In the context of combined recirculation, the recirculation ratio (X₃) was defined as the volumetric percentage of total recirculated production wastewater—comprising both SSW and FBW—relative to the raw water volume. As reported in Section 3.1, the results of single‐stream recirculation experiments indicated that the optimal recirculation ratios for SSW and FBW were approximately 4% and 6%, respectively. To explore potential synergistic effects and establish appropriate optimization ranges for combined recirculation, the center point (0 level) of the RSM design was set at a total recirculation ratio of 6%, with 2% and 10% designated as the low (−1) and high (+1) levels, respectively.

During combined recirculation trials, the individual recirculation volumes for each wastewater stream were determined by halving the total target recirculation ratio and then scaling according to the respective proportions specified in the experimental design. The complete experimental design matrix for the combined recirculation experiments is presented in Table 2.

TABLE 2.

Experimental factor levels.

Variable	Factor	Factor level
		−1	0	1
Recirculation point	X1	−1	0	1
Coagulant dosage (mg/L)	X2	10	15	20
Recirculation ratio	X3	2	6	10

Note: Recirculation point −1 denotes predosage recirculation; recirculation point 0 denotes recirculation prior to postchemical dosing but premixing; recirculation point 1 denotes recirculation postchemical mixing.

3. Results and Discussion

3.1. Effect of Individual Recirculation on Effluent Water Quality

3.1.1. Water Quality Parameters

The effects of different wastewater treatment residuals (WTRs) on effluent water quality under separate recirculation conditions are presented in Figure 1. Turbidity, a critical operational parameter in drinking water treatment, serves as a key indicator of coagulation‐sedimentation performance. As shown in Figure 1a, when the four types of WTR were recirculated individually, effluent turbidity initially decreased and subsequently increased with rising recirculation ratios across all scenarios. Notably, all recirculation conditions resulted in superior treatment outcomes compared to the nonrecirculation baseline. This finding aligns with previous research, which has demonstrated that residual coagulants present in WTR enhance flocculation by providing additional nucleation sites, thereby improving overall coagulation efficacy (Sillanpää et al. 2018). The optimal recirculation ratios for SSW and FBS were determined to be 4%, while SSS and FBW achieved optimal performance at 6%. The relative contribution of each recirculation scenario to turbidity removal efficiency followed the order: SSW > SSS > FBW > FBS.

FIGURE 1.

Effect of different WTRs on effluent water quality during separate recirculation.

In contrast, no substantial improvement in ammonia nitrogen (NH₃‐N) removal was observed under any of the recirculation conditions, although all four scenarios exhibited variations in removal efficiency as a function of recirculation ratio (Figure 1b). Notably, SSW and FBW demonstrated superior recirculation effects compared to their respective supernatants (SSS and FBS). This difference may be attributed to the adsorption of NH₃‐N by larger floc particles present in the raw waste streams, whereas the supernatant samples exhibited limited adsorptive capacity, occasionally resulting in negative removal rates (Matilainen et al. 2010). The optimal recirculation ratios for NH₃‐N removal were 8% for SSW and FBW, and 4% for SSS and FBS.

The four recirculation scenarios also exhibited marked differences in organic matter removal. As illustrated in Figure 1c, FBW recirculation achieved the highest CODMn removal efficiency, which remained relatively stable across increasing recirculation ratios. FBS recirculation ranked second in CODMn removal performance but demonstrated greater sensitivity to variations in recirculation ratio. In contrast, under SSW and SSS recirculation, CODMn removal efficiency declined with increasing recirculation ratio, eventually reaching negative removal rates. This phenomenon is likely attributable to the elevated concentrations of both particulate and dissolved organic matter present in SSW and SSS (see Table 1), combined with the limited efficacy of conventional coagulation‐sedimentation processes in removing certain organic compounds (Zhou et al. 2012).

Notably, all four recirculation methods exhibited highly similar UV₂₅₄ removal behavior, characterized by an initial increase followed by a decrease with increasing recirculation ratio, with optimal removal achieved at a 4% recirculation ratio (Figure 1d). The relative contribution of each recirculation condition to UV₂₅₄ removal followed the order: SSS > FBS > SSW > FBW. Importantly, the two supernatant conditions (SSS and FBS) achieved higher UV₂₅₄ removal rates than their corresponding raw waste streams (SSW and FBW), indicating lower concentrations of aromatic dissolved organic matter in the supernatants.

3.1.2. Al, Fe, and Mn Parameters

As shown in Figure 2, the four recirculation scenarios exhibited distinct effects on effluent concentrations of aluminum (Al), iron (Fe), and manganese (Mn) with increasing recirculation ratio. Notably, all four conditions adversely influenced Mn removal, with SSW and SSS recirculation posing the greatest risk of Mn enrichment. These findings highlight the need for careful consideration of Mn accumulation when implementing production wastewater recirculation.

FIGURE 2.

Effect of four different recirculation conditions on Al‐Fe‐Mn.

For iron and aluminum, the results indicate that recirculation ratios for FBW and FBS should be maintained below 4% to minimize adverse effects on effluent quality. In contrast, the use of SSW and SSS should be avoided where feasible, particularly in water treatment plants treating raw water with elevated iron concentrations.

In summary, for raw water sources characterized by high concentrations of aluminum, iron, or manganese, the recirculation of SSW and SSS should be approached with extreme caution. Where FBW or FBS recirculation is employed, the recirculation ratio should be carefully controlled and maintained below 4% to safeguard effluent quality.

3.2. Optimization of Combined Recirculation Parameters Based on RSM

The single‐factor experiments described in Section 3.1, conducted under fixed coagulant dosage and recirculation point conditions, demonstrated that the recirculation ratio significantly influences pollutant removal efficiency and established its optimal operational range. However, in full‐scale water treatment processes, complex interactions exist among multiple operational parameters, including recirculation point, coagulant dosage, and recirculation ratio. To elucidate these interactions and identify globally optimal process conditions, the present study employed RSM to systematically investigate the combined effects of three key factors—recirculation point (X₁), coagulant dosage (X₂), and recirculation ratio (X₃)—on effluent quality during combined recirculation of SSW and FBW.

3.2.1. Combined Recirculation SSW and FBW

To optimize turbidity removal efficiency, it was hypothesized that combined recirculation of SSW and FBW would yield synergistic benefits, given that single‐stream recirculation achieved optimal performance at ratios of 4% and 6%, respectively. Accordingly, in the combined recirculation experiments, the original optimal ratios were halved—resulting in initial mixing proportions of 2% for SSW and 3% for FBW—to serve as the baseline for further optimization. RSM was subsequently employed to systematically optimize the combined recirculation process. The experimental design matrix and corresponding results are presented in Table 3.

TABLE 3.

Response surface design and results.

Serial number	X1: recirculation point	X2: coagulant dosage	X3: recirculation ratio	Turbidity removal rate (%)		UV254 removal rate (%)
				Observed	Predicted	Observed	Predicted
1	0	20	10	90.64	90.5	56.34	56.37
2	1	15	10	87.72	88.07	53.52	54.46
3	1	15	2	90	89.59	59.15	60.8
4	0	15	6	94.23	94.69	60.56	62.56
5	0	15	6	94.29	94.69	63.38	62.56
6	0	10	2	92.12	91.32	59.15	57.91
7	0	15	6	94.1	94.69	66.2	62.56
8	1	20	6	90.06	88.95	61.97	62.36
9	−1	15	2	92.82	92.59	60.56	60.84
10	0	10	6	87.88	92.26	57.75	59.64
11	1	15	6	94.55	91.48	64.79	61.82
12	−1	10	6	91.09	92.17	60.56	60.44
13	−1	15	10	91.54	91.99	57.75	57.34
14	0	15	6	96.28	94.69	60.56	62.56
15	0	20	2	87.69	89.2	61.97	61.29
16	0	10	10	89.36	87.9	53.52	52.99
17	−1	20	6	94.29	93.09	63.38	63.66

Multivariate regression analysis was performed on the experimental data obtained for Combined Recirculation Method 1 using Design‐Expert software. The regression analysis yielded two quadratic models for the turbidity removal rate (Y1) and the UV₂₅₄ removal rate (Y2), which were subsequently plotted against the recirculation point (X1), the dosage (X2), and the recirculation ratio (X3). The plots are presented below:

$$\begin{array}{c}\text{Y1}=94.66-1.73{\text{X}}_1+0.12{\text{X}}_2-0.53{\mathrm{X}}_3-0.34{\text{X}}_1{\text{X}}_2-0.23{\text{X}}_1{\mathrm{X}}_3\hfill \\ +1.18{\text{X}}_2{\mathrm{X}}_3-1.48{{\text{X}}_1}^2-2.31{{\text{X}}_2}^2-2.65{{\mathrm{X}}_3}^2\hfill \end{array}$$

$$\begin{array}{c}\text{Y2}=62.56-0.73{\text{X}}_1+1.69{\text{X}}_2-2.46{\mathrm{X}}_3+0.08{\text{X}}_1{\text{X}}_2-0.71{\text{X}}_1{\mathrm{X}}_3\hfill \\ -0.0135{{\text{X}}_1}^2-1.23{{\text{X}}_2}^2-4.19{{\mathrm{X}}_3}^2\hfill \end{array}$$

As illustrated in Table 4, the results of the analysis of variance (ANOVA) are presented for the statistical significance of the models. This includes the coefficient of determination (R ²), sum of squares, mean square, F value, and probability (p value). When the p value is below the significance level of 0.05, the model is deemed statistically significant, indicating that at least one independent variable in the model exhibits a significant correlation with the dependent variable (Sibiya et al. 2022). The ANOVA for turbidity (Y1) and UV₂₅₄ (Y2) removal rates demonstrates that both models exhibit p values below 0.05, thereby confirming their statistical significance. The signal‐to‐noise ratios for these models are 48.7894 and 7.0616, respectively. It is evident that both values exceed the acceptable threshold of 4, indicating sufficient signal strength for navigating the design space.

TABLE 4.

Analysis of variance statistics for the reaction model.

Response	Turbidity		UV254
p values		< 0.0001	0.0303
F values		291.41	4.48
Mean of squares		9.60	17.38
Sum of squares errors		86.43	156.38
Standard deviation		0.1815	1.97
Mean		91.62	60.01
Coefficient of variance (CV, %)		0.1981	3.28
Coefficient of determination		0.9973	0.8521
Adjusted R 2		0.9939	0.6620
Predicted R 2		0.9571	0.1922
Adequate precision		48.7894	7.0616

It should be noted, however, that the predicted R ² values for UV₂₅₄ removal are considerably lower (0.1922 for the SSW + FBW model and 0.2038 for the SSS + FBS model) than the generally acceptable threshold of 0.5 (Gadekar and Ahammed 2019). While both models remain statistically significant (p < 0.05) and provide valuable insights into the directional effects of operational parameters, their capacity to accurately predict absolute UV₂₅₄ removal rates under untested conditions is limited. Therefore, the optimal conditions identified for UV₂₅₄ removal should be interpreted as indicative trends rather than precise optima, and further validation experiments covering a broader range of operational scenarios are recommended before generalizing the model predictions.

Figure 3 presents the response surface plots for turbidity and UV₂₅₄ removal efficiency under combined recirculation conditions, illustrating the interactions among recirculation point, coagulant dosage, and recirculation ratio.

FIGURE 3.

Response surface plots showing the interactive effects of recirculation point, coagulant dosage, and recirculation ratio on turbidity and UV₂₅₄ removal efficiency under combined recirculation of SSW and FBW.

The elliptical contour observed for the interaction between dosage and recirculation ratio indicates a significant combined effect on turbidity removal. Further analysis confirms that the recirculation ratio is the dominant factor influencing turbidity removal, consistent with the findings of Tong et al. (2025). The contour lines within the response surface diagram indicate that the maximum turbidity removal rate occurs at the center of the smallest ellipse, corresponding to the optimal conditions: a coagulant dosage of 16.06 mg/L and a recirculation ratio of 6.52%. Under these conditions, the maximum turbidity removal rate reached 94.86%.

Regarding UV₂₅₄ removal, the elliptical contour observed for the interaction between dosage and recirculation point indicates a significant combined effect on the removal of UV₂₅₄‐absorbing organic compounds. Notably, the response surface for predosage recirculation encompassed the largest high‐efficiency removal region, confirming that recirculation point is a critical factor influencing UV₂₅₄ removal. The peak UV₂₅₄ removal efficiency was achieved under predosage recirculation conditions, with a coagulant dosage of 18.27 mg/L and a recirculation ratio of 5.16%, yielding a maximum UV₂₅₄ removal rate of 63.98%.

The findings of the comprehensive response surface analysis suggest that, in the context of combined recirculation of SSW and FBW, the recirculation ratio exerts a more significant influence on turbidity and UV₂₅₄ removal rates in comparison to the dosage. Therefore, controlling the total recirculation ratio is critical for process optimization. It is evident that, under the optimal operating parameters for both indicators, namely, a recirculation point predosage, a dosage of 16.84 mg/L, and a combined recirculation ratio of 5.38%—turbidity and UV₂₅₄ removal rates can reach 94.69% and 63.87%, respectively.

3.2.2. Combined Recirculation of SSS and FBS

Regarding UV₂₅₄ removal efficiency, the optimal single‐stream recirculation conditions for SSW and FBW were achieved at a recirculation ratio of 4% for their respective supernatants (SSS and FBS). Accordingly, in the combined recirculation experiments involving SSS and FBS, the recirculation ratio was set to 4% as the baseline for further optimization. The RSM experimental design and corresponding results for combined SSS and FBS recirculation are presented in Table 5.

TABLE 5.

Response surface design and results.

Serial number	X1: recirculation point	X2: coagulant dosage	X3: recirculation ratio	Turbidity removal rate (%)		UV254 removal rate (%)
				Observed	Predicted	Observed	Predicted
1	0	15	6	95.58	96.13	60.56	67.32
2	1	15	10	90.78	90.4	46.48	41.54
3	0	20	2	94.29	93.96	45.07	40.84
4	0	20	10	91.43	91.76	25.35	30.98
5	–1	10	6	95.06	95.01	40.85	41.55
6	–1	15	10	93.83	93.54	52.11	47.18
7	–1	20	6	95.58	95.53	54.93	54.23
8	1	20	6	94.81	94.87	50.7	49.99
9	0	10	2	93.57	93.24	32.39	26.76
10	0	15	6	95.52	96.13	71.83	67.32
11	–1	15	2	94.68	95.06	50.7	55.64
12	0	15	6	96.36	96.13	70.42	67.32
13	1	15	2	93.7	94	46.48	51.4
14	0	10	10	90	90.32	14.08	18.3
15	1	10	6	91.43	91.47	35.21	35.91
16	0	15	6	96.43	96.13	69.01	67.32
17	0	15	6	96.75	96.13	64.79	67.32

Multivariate regression analysis of the combined recovery experimental data using software yielded quadratic regression models for turbidity removal rate (Y1) and UV₂₅₄ removal rate (Y2) versus recovery point (X1), coagulant dosage (X2), and recirculation ratio (X3).

$$\begin{array}{c}\text{Y1}=\mathrm{96.131.05}–{\text{X}}_1+\mathrm{0.541.28}{\text{X}}_2–{\mathrm{X}}_3+\mathrm{0.720.52}{\text{X}}_1{\text{X}}_2–{\text{X}}_1{\mathrm{X}}_3\hfill \\ +\mathrm{0.180.491.422.39}{\text{X}}_2{\mathrm{X}}_3–{{\text{X}}_1}^2–{{\text{X}}_2}^2–{{\mathrm{X}}_3}^2+0.44{{\text{X}}_1}^2{\text{X}}_2\hfill \end{array}$$

$$\begin{array}{c}\mathrm{Y}2=\mathrm{67.322.47}–{\text{X}}_1+\mathrm{6.694.58}{\text{X}}_2–{\mathrm{X}}_3\hfill \\ +\mathrm{0.350.350.351.0920.8117.29}{\text{X}}_1{\text{X}}_2–{\text{X}}_1{\mathrm{X}}_3\hfill \\ –{\text{X}}_2{\mathrm{X}}_3–{{\text{X}}_1}^2–{{\text{X}}_2}^2–{{\mathrm{X}}_3}^2\hfill \end{array}$$

ANOVA was performed to evaluate the statistical significance of the quadratic regression models for turbidity removal (Y₁) and UV₂₅₄ removal (Y₂) under combined SSS and FBS recirculation. The results include the coefficient of determination (R ²), sum of squares, mean square, F value, and probability (p value). A p value below the 0.05 significance level indicates that the model is statistically significant, meaning that at least one independent variable exhibits a meaningful relationship with the response variable. As shown in Table 6, both models yielded p values below 0.05, confirming their statistical significance. The signal‐to‐noise ratios, which measure the adequacy of model discrimination, were 12.1683 for turbidity removal and 10.0457 for UV₂₅₄ removal. Both values substantially exceed the minimum acceptable threshold of 4, indicating that the models possess sufficient signal strength to reliably navigate the design space.

TABLE 6.

Analysis of variance statistics for the reaction model.

Response	Turbidity	UV254
p values	0.0009	0.0026
F values	18.90	10.62
Mean of squares	6.63	429.98
Sum of squares errors	1.20	84.90
Standard deviation	0.5921	6.36
Mean	94.11	48.88
Coefficient of variance (CV, %)	0.6292	13.01
Coefficient of determination	0.9692	0.9318
Adjusted R 2	0.9179	0.8441
Predicted R 2	0.4450	0.2038
Adequate precision	12.1683	10.0457

Figure 4 presents the response surface plots illustrating the interactions among recirculation point, coagulant dosage, and recirculation ratio on turbidity and UV₂₅₄ removal efficiency under combined recirculation of SSS and FBS.

FIGURE 4.

Response surface plots showing the interactive effects of recirculation point, coagulant dosage, and recirculation ratio on turbidity and UV₂₅₄ removal efficiency under combined recirculation of SSS and FBS.

The elliptical contour observed for the interaction between dosage and recirculation ratio indicates a significant synergistic effect on turbidity removal efficiency. Subsequent analysis confirmed that the recirculation ratio is the primary factor influencing turbidity removal under these conditions. The contour lines within the response surface diagram demonstrate that maximum turbidity removal occurs at the center of the smallest ellipse, corresponding to the optimal conditions: predosage recirculation, a coagulant dosage of 15.37 mg/L, and a recirculation ratio of 5.53%. Under these conditions, the maximum turbidity removal efficiency reached 96.76%.

Regarding UV₂₅₄ removal, the elliptical contour observed for the interaction between dosage and recirculation point indicates a significant combined effect on the removal of UV₂₅₄‐absorbing organic compounds. Notably, the response surface for predosage recirculation encompassed the largest high‐efficiency removal region, further validating that recirculation point is a critical factor influencing UV₂₅₄ removal. The peak UV₂₅₄ removal efficiency was achieved under predosage recirculation conditions, with a coagulant dosage of 15.77 mg/L and a recirculation ratio of 5.51%, yielding a maximum UV₂₅₄ removal rate of 69.44%.

The comprehensive response surface analysis indicates that the recirculation ratio exerts a more pronounced influence on both turbidity and UV₂₅₄ removal efficiency than coagulant dosage under combined SSS and FBS recirculation. Under the globally optimized conditions—specifically, predosage recirculation, a coagulant dosage of 15.73 mg/L, and a combined recirculation ratio of 5.50%—turbidity and UV₂₅₄ removal efficiencies reached 96.75% and 69.44%, respectively.

3.3. Multiresponse Optimization

As expected, both turbidity removal efficiency and UV₂₅₄ removal efficiency are distinct responses, each optimized under separate conditions. Consequently, trade‐offs between these response conditions are desirable (Khumalo et al. 2023). It is possible to derive a trade‐off regression equation that accounts for both turbidity and UV₂₅₄ removal efficiency from the feasibility function. The feasibility function equation for the combined recirculation SSW and FBW is as follows:

$$\begin{array}{c}Y=95.1+0.120{\text{X}}_1–.{\text{1X}}_2–1.\text{5}{\mathrm{X}}_3–0.01{\text{X}}_1{\text{X}}_2\hfill \\ +\mathrm{0.150.05}{\text{X}}_1{\mathrm{X}}_3–{\text{X}}_2{\mathrm{X}}_3+0.05{{\text{X}}_1}^2+0.00{{\text{2X}}_2}^2+0.\text{8}{{\mathrm{X}}_3}^2\hfill \end{array}$$

The regression equation was utilized to calculate the optimal process conditions, which included the recirculation point predosage, the coagulant dosage of 16.84 mg/L, and the recirculation ratio of 5.38%. In these conditions, the turbidity removal rate and the UV₂₅₄ removal rate were found to be 94.69% and 63.87%, respectively. Three validation experiments were conducted under compromise conditions, yielding average removal efficiencies of 93.57% and 62.98% across three trials. The results agreed well with model predictions, with the predictions made by the model.

The feasibility function equation for the SSS and FBS combination is as follows:

$$\begin{array}{c}Y=96.75+\mathrm{0.050.482.430.00}{\text{X}}_1–{\text{X}}_2–{\mathrm{X}}_3–{\text{1X}}_1{\text{X}}_2+0.01{\text{2X}}_1{\mathrm{X}}_3\hfill \\ +0.00{\text{2X}}_2{\mathrm{X}}_3+0.03{{\text{8X}}_1}^2+0.01{{\text{5X}}_2}^2+1.28{{\mathrm{X}}_3}^2\hfill \end{array}$$

The regression equation was utilized to calculate the optimal process conditions, which included the recirculation point predosage, the coagulant dosage of 15.77 mg/L, and the recirculation ratio of 5.51%. At this juncture, the turbidity removal rate and the UV₂₅₄ removal rate were 96.75% and 69.44%, respectively. Three validation experiments were conducted under compromise conditions, yielding average removal rates of 96.24% and 70.05% across two trials. The results agreed well with model predictions made by the model.

Finally, it is important to recognize that the present study was conducted at laboratory scale using jar tests with simulated recirculation conditions. While the experimental design captured key operational variables and interactions, factors such as continuous long‐term operation, variations in raw water quality across seasons, and potential accumulation of trace contaminants (e.g., Mn, Al, or disinfection byproduct precursors) were not explicitly addressed. Therefore, the proposed combined recirculation strategy should be validated in pilot‐ or full‐scale systems before being recommended for routine implementation in drinking water treatment plants.

4. Conclusions

1. Single‐stream recirculation experiments demonstrated that direct recirculation of production wastewater (SSW and FBW) at ratios of 4%–6% maximized turbidity removal. In contrast, recirculation of their supernatants (SSS and FBS) at a 4% ratio yielded superior removal of UV₂₅₄ and CODMn. However, caution is warranted regarding the risk of manganese (Mn) enrichment associated with SSW and SSS recirculation; for raw water sources with elevated Mn levels, recirculation ratios should be maintained below 4% to safeguard effluent quality.

2. Response surface optimization revealed that the recirculation ratio is the predominant factor governing pollutant removal efficiency in combined recirculation processes. For combined SSW and FBW recirculation, the optimal conditions were predosage recirculation, a coagulant dosage of 16.84 mg/L, and a recirculation ratio of 5.38%, achieving turbidity and UV₂₅₄ removal rates of 94.69% and 63.87%, respectively. For combined SSS and FBS recirculation, the optimal conditions were predosage recirculation, a dosage of 15.77 mg/L, and a recirculation ratio of 5.51%, yielding enhanced removal rates of 96.73% for turbidity and 70.05% for UV₂₅₄. Notably, the SSS and FBS combination outperformed the SSW and FBW combination in removing both turbidity and UV₂₅₄.

3. Under the tested laboratory conditions, the combined recirculation process showed potential for coagulant savings of approximately 15%–20% (from 15 to 12–13 mg/L) while maintaining turbidity removal above 90%. This optimization strategy may offer a technically feasible and energy‐efficient option for treating low‐turbidity SWs, particularly in plants seeking to reduce coagulant consumption.

Author Contributions

Zhengong Tong: methodology, conceptualization, validation, formal analysis, project administration, resources, supervision, writing – original draft, writing – review and editing, funding acquisition. Ruikang Hu: software, validation, investigation, writing – review and editing. Zhicheng Xi: data curation, validation, formal analysis, visualization, writing – original draft, investigation, software, writing – review and editing. Tongtong Zhang: validation, investigation, supervision. Zhewei Chen: investigation, validation, supervision. Shuqi Wang: investigation, validation, supervision.

Funding

This work was supported by the National Natural Science Foundation of China (52060006, 42001111, and 52560002).

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

Data will be made available on request.