Enhanced Treatment of Decentralized Domestic Sewage Using Gravity-Flow Multi-Soil-Layering Systems Coupled with Iron-Carbon Microelectrolysis

Abstract

Soil-based decentralized treatment technology has become increasingly popular as an ideal solution for water pollution control in rural areas. It is very necessary to optimize the removal mechanisms and performance of such technologies on rural domestic sewage treatment. This was the first study of a gravity-flow multi-soil-layering (MSL) system coupled with iron-carbon microelectrolysis (ICM). Influent COD/TN (C/N) ratio and bottommost soil mixture block (SMB) submersion were selected as the operating factors relevant to the ICM in MSL systems. Such two key factors were investigated in the factorial experiment. The removal efficiencies of COD, TP, NH₃-N, NO₃⁻-N, and TN could be reached up to 96.3, 100, 95.4, 93.8, and 79.6%, respectively. Different levels of factors could comprehensively drive the performance variation. The factorial analysis indicated that the bottommost SMB submersion had the most significant and dominant negative effects on aerobic processes. The ideal TP removal attributed to the presence of the bottommost SMB submersion. It played the dominant role for the bottommost SMB submersion in facilitating an electrochemical reaction through the ICM. Zero-valent iron or ferrous ions could be transformed to final ferric ions more efficiently during the period of the ICM reactions. The ICM could promote the capability of a SMB for removing nutrients in sewage, especially provide electron donors to denitrifying bacteria in MSL systems. However, there were non-significant effects of the influent C/N ratio on the removal performance of MSL systems. This study can help enrich the pollutant removal mechanisms in MSL systems.

1. Introduction

Along with the development of the rural ecotourism economy, there is a large amount of rural domestic sewage discharged disorderly. The increasing pollution load on regional water poses serious risks to environmental security and to the health of local people [1]. Decentralized domestic sewage has become one of the main pathways of the organic and nutrient pollution finding its way into the rural water environment [2,3]. It also caused potential eutrophication and even the nitrate contamination of groundwater [4]. This issue is of growing concern in remote rural areas and small communities all over the world [5,6].

During the last two decades, a number of studies had employed emerging soil-based technologies. For instance, natural soil aquifers and constructed wetlands (CWs), as the ecological treatment options for facilitating safe water reuse in remote rural areas, instead of inefficient septic tanks [7,8]. However, some studies had also identified the inadequacies of CWs used in rural areas, such as the large land space requirement, low hydraulic loading rate, and regional clogging [9]. Municipal centralized sewage treatment with the necessary expensive installations, professional skills, and routine maintenance [5,10]. This is significantly uneconomical for remote rural areas and small communities with dispersed households and a low population density. Therefore, an alternative eco-technology with more advanced characteristics is much needed in rural areas, particularly for the remote areas of developing countries. At present, the soil-based, modularized, and decentralized treatment technology has become increasingly popular as an ideal solution for water pollution control in such areas [11,12,13].

The gravity-flow multi-soil-layering (MSL) system is a typical soil-based and eco-friendly technology for treating decentralized rural sewage [9,10]. It has been applied over the world and is now widely attracting attention as well [14,15]. There are two types of modules that were installed in the MSL system (permeable layer (PL), soil mixture block (SMB)) to significantly facilitate the aerobic and anaerobic conditions, respectively [16]. Moreover, the MSL system has various advantageous features, such as an intensive space utilization, a net zero energy consumption, smooth flow, easy maintenance, good aesthetic, and no odor [9,15]. In the traditional sense, several complicated processes, such as complexation, precipitation, filtration, adsorption, and biodegradation are involved in the MSL system [17]. The microbial process is seen to be the primary driver for removing organics and nitrogen [14,18]. Chemical complexation and precipitation are considered as the main mechanisms for phosphorus removal [19].

There were previous numbers of MSL-based studies and the research findings were inspiring [1,9]. The ICM had been widely used in the treatment of sewage from the dyeing, breeding, and pharmaceutical industries [20,21,22,23,24]. However, few physicochemical mechanisms related to iron-carbon microelectrolysis (ICM) for the removal of pollutants in the MSL systems were followed with interest and emphasized. The performance of MSL systems, coupled with the ICM, on decentralized domestic sewage treatment have not been studied yet. Furthermore, the interactive effect of key operating factors with different levels on the process of microelectrolysis in the MSL systems is not clear. It is thus of interest to explore such potential electrochemical mechanisms of the ICM in the MSL systems, with the reactive media condition, based on a zero-valent iron and carbon-based materials.

A factorial analysis is advantageous for tackling the statistical relationships between a targeted pollutant removal and other dynamical conditions and parameters [19,25]. Based on the two-level factorial design experiment, this study aims to (i) analyze the effects of the influent C/N ratio and a bottommost SMB submersion on the performance of the ICM-based MSL systems, and (ii) explore the interactions among two key operating factors in sewage treatment processes. This study investigated and explained the microelectrolysis mechanisms for treating domestic sewage through MSL systems from the factorial perspective for the first time as well. The results can help enrich the removal mechanisms of pollutants in MSL systems, and provide the theoretical foundations for the optimal design and operation of on-site application.

2. Materials and Methods

2.1. ICM-Based MSL Systems

The experimental MSL systems, as shown in Figure 1, were made of acrylic materials, and were designed with L × W × H dimensions of 52 × 10 × 80 cm. The two holes at the top and one hole at the bottom of each system were designed for attaching the influent distribution pipe and effluent outlet pipe, respectively. The bottommost sampling port was constructed using an 8 cm length of PVC pipe connected to the side of each MSL system.

The fresh soils used in the lab-scale experiment were collected from the surface (5 to 15 cm) of a garden field. The soil pre-samples were dispersed, homogenized, and sieved (2 mm mesh), to remove plant rootlets and small stones. Polymer poly-butylene (Showa Denko Co., Ltd., Tokyo, Japan) has a good biocompatibility and microbial absorbability, which was added to the SMBs. It serves as a high-efficiency and slow-release carbon source material for long-term operation in soil-based systems. Surface soil, grain-type PBS, iron powder (Fe), and biochar powder (C) (Shandong Tairan Biological Engineering Co., Ltd., Dongying, China) were packaged with a dry-weight ratio of 7:1:1:1 in each SMB jute bag, which was the same as the classic ratio, previously [8,17]. Furthermore, a 10 g microbial agent of microtherm activated sludge (MAS) (Bio-Form Co., Ltd., Guangzhou, China), which was mainly composed of Bacillus, Enterococcus, and Blastocystis, was also added in each SMB jute bag for enhancing the species and the abundance of functional microorganisms in the MSL systems.

The better pollutant removal efficiencies could be obtained along with a more effective connection between the porous media and sewage [26]. The relatively large top areas for the SMBs served to better intensify the sewage distribution and its contact with materials in the MSL systems [19,27]. Therefore, there were two L × W × H dimension patterns of SMB with 23 × 10 × 6 and 16 × 10 × 6 cm in this experiment. The horizontal distance between the SMBs was 2 cm, and each PL between the SMBs was 8 cm in height for the MSL system with the unique brick wall pattern. Biological ceramsite (Wanyuan Water Purification Materials Co., Ltd., Gongyi, China) with the homogeneous dimensions of 8 mm, as the environment-friendly and low-cost adsorbing material, was applied in the PLs. Prior to the effluent outlet of the MSL system, the system was configured with fine grid plastic-steel plates featuring 2 mm holes for isolating the bottom biological ceramsite of the PLs and impurities.

2.2. Factorial Design Experiment

Several studies indicated that the ICM was advantageous for nitrate removal by reducing the dependence of denitrification on the traditional carbon sources [28,29]. Furthermore, the electrochemical efficiency of the ICM could be influenced by the redox conditions and the water flow situation [30], which were mainly corresponding to the ORP variation and the submersion, or not, in the MSL systems, respectively. Thus, the most important two operating factors, the influent COD/TN (C/N) ratio (A) and the bottommost SMB submersion (B), were considered in this study. The experimental scheme, based on a 2-level (2²) factorial design, was shown in Figure 2. The high and low levels of the operating factors were coded +1 and −1, respectively. The actual values of the high and low levels for factor A were 5:1 and 3:1, and the actual values of the high and low levels for factor B were 20 cm (effluent outlet #1) and 2 cm (effluent outlet #2) of the effluent height, respectively.

2.3. Synthetic Sewage

The physicochemical parameters of the water including pH, dissolved oxygen (DO), oxidation-reduction potential (ORP), and electrical conductivity (EC) of the effluent, as well as the water quality indices, including the chemical oxygen demand (COD), total phosphorus (TP), ammonia nitrogen (NH₃-N), nitrate nitrogen (NO₃⁻-N), and the total nitrogen (TN), were periodically detected for evaluating the performance of the MSL systems. According to the factorial design scheme and the universal features of the raw domestic sewage in rural areas [3,9,31], the quantity of chemicals for the preparation used throughout this lab-scale experiment, and the water quality of the synthetic sewage, were determined, as shown in

Table 1. The quantity of chemicals for preparing the synthetic sewage and the water quality under different scenarios.

Chemical Agent	Amount (mg/L)	Water Quality Index	Concentration (mg/L)
			Sewage 1	Sewage 2
Potassium biphthalate	0.13/0.06	COD	400	240
Glucose	0.09/0.05	TP	5
Monopotassium phosphate	0.04	NH3-N	65
Ammonium sulfate	0.37	NO3−-N	13
Carbamide	0.12	TN	80
Sodium nitrate	0.06	C/N ratio	5:1	3:1

. The initial pH, DO, ORP, and EC of the synthetic sewage were 7.13, 8.34 mg/L, 196 mV, and 1.77 uS/cm, respectively.

2.4. Operation of the Lab-Scale Experiment

There were four MSL systems which were denoted as MSL1 to MSL4 (Figure 2), two 120 L plastic barrels for the synthetic sewage supply, and a centralized effluent collection pipe were used in experiment. During the experimental period, the synthetic sewage was fed continuously into the MSL systems by peristaltic pumps (Kamoer Fluid Tech Co., Ltd., Shanghai, China), through three water droppers on the upmost silicone tubing (inner diameter 1.5 mm, outside diameter 4.0 mm). During the experimental period, the temperatures of all MSL systems were kept at 20 ± 1 °C. Furthermore, the hydraulic loading rate (HLR) was set at 300 L/(m²d) for all MSL systems. Throughout the 80-day operational period, the raw effluent from the sampling outlet of each MSL system was collected at 10-day intervals for the purpose of monitoring the water quality.

2.5. Analytical Methods

All of the water physicochemical parameters and water quality parameters were measured immediately after sampling and according to the standard methods for the examination of water and sewage [32]. COD was digested by a Hach DRB/200 digester (Loveland, CO, USA) at 150 °C for 120 min and determined by a Hach DR/890 spectrophotometer (Loveland, CO, USA), according to the standard colorimetric determination method 5520D with Hach 21259-25 medicament. TP was digested by Hach DRB/200 at 150 °C for 30 min and detected by Hach DR/890, according to the standard molybdovanadate with acid persulfate digestion method 4500 B-C (high range) with Hach 27672-45 medicament. TN was digested using Hach DRB/200 at 105 °C for 30 min and determined by Hach DR/890, according to the standard persulfate digestion method 4500 P-J (high range) using Hach 27141-00 medicament. NH₃-N and NO₃⁻-N were detected by Hach DR/890, according to the standard salicylate method 4500-NH₃ (low range) using Hach 26069-45 medicament and the standard cadmium reduction method using Hach 2106-69 medicament, respectively. pH, DO, ORP, and EC were measured with a S40 pocket pH tester (BANTE, Shanghai, China), HI 98186 dissolved oxygen meter (HANNA, Italy), a S20 pocket ORP tester (BANTE, Shanghai, China), and HI 98186 conductivity meter (HANNA, Rome, Italy), respectively. Each 500 mL effluent sample from the MSL system was filtered through medium speed filter papers to remove suspended impurities. pH, DO, ORP, and EC were measured in unfiltered samples directly after collecting the effluent samples. The detailed methods can also be referred to our previous study [18,19,27].

2.6. Data Analysis

Design-expert software (Stat-Ease, Inc., Minneapolis, MN, USA) was used for factorial analysis through the results of a one-way analysis of variance (ANOVA). The OriginPro software (OriginLab Corporation, Northampton, MA, USA) was used for drafting.

3. Results and Discussion

3.1. Variation of the pH, DO, and ORP in the MSL Systems

The variations of the physicochemical parameters, including the pH value, DO concentration, and ORP value of the effluent are shown in Figure 3a–c, respectively. The variation of the pH value in MSL1 was obviously increased from 7.32 to 8.38 during days 0 to 80. In the same period, the pH values in MSL2 and MSL3 were slightly increased from 7.25 to 7.43 and 7.18 to 7.39, respectively. However, the pH variation was barely changed in MSL4 with a range from 7.17 to 7.24. The results showed that the higher influent C/N ratio, the larger the increase of the pH value. It was mainly attributed to the more efficient denitrification with the acidity depletion of sewage, especially in the MSL systems with the bottommost SMB submersion (anaerobic condition) [8,14]. There were inapparent differences in the pH variation, even the final pH value between MSL2 and MSL3, due to the relatively sufficient oxygen concentration (aerobic condition) and the low C/N ratio, respectively.

During days 0 to 30, there were significant decreases of the DO concentration in the MSL systems at the initial operation stage, due to the intensive oxygen consumption for activating the microorganisms and the chemical oxidized reactions of the ICM [33]. In detail, the DO concentrations deceased from 3.65 to 3.01 mg/L for MSL1, 6.67 to 5.54 mg/L for MSL2, 3.92 to 2.99 mg/L for MSL3, and 6.24 to 5.49 mg/L for MSL4, respectively. Following day 30, the DO concentrations in MSL1 and MSL3 were stable at the average (maximum) concentration of 3.29 (3.41) and 3.27 mg/L (3.22 mg/L), respectively. However, the DO concentrations in MSL2 and MSL4 were significantly increased to the final 7.13 and 6.98 mg/L, respectively. These were mainly impacted by the bottommost SMB submersion, or by the absence of a bottommost SMB submersion, in the four MSL systems, which were divided into semi-closed and unobstructed types.

MSL2 and MSL4 showed relatively higher average ORP values of 123.5 and 103.5 mV during day 0 to 20, respectively, which showed the good oxidation condition (more than 100 mV) [19]. Due to MSL1 and MSL3 with the long-term bottommost SMB submersion, the ORP values of these two systems continued to fall throughout the experiment and ranged from an initial 79 and 83 mV to final ORP values of 35 and 38 mV, respectively. It indicated that the final strongest reductive conditions were found in the two such MSL systems. Although there was a slight fluctuation of the ORP values from 72 to 82 mV, MSL4 performed the most relatively stable and the least weakly reducing environment, due to the combined effects of not having a bottommost SMB submersion and fewer organics in the influent were oxidized [8,17]. MSL2 also showed a stable but higher ORP level at an average of 115.3 mV, which maintained the relatively predominant oxidation environment after day 30. In general, the variation of the ORP trailed behind that of DO, in the MSL systems. Rather than the DO concentration, the ORP value could be more intuitive to reflect the redox properties in the MSL system [27]. The ORP value could be adjusted by various microcosmic mechanisms underlying the physicochemical and biochemical processes [1]. Along with the gradually matured processes in the PLs and the SMBs, there was a relative stabilization of the DO concentrations and the ORP values.

3.2. ICM in the SMBs and the EC Variation in the MSL Systems

Based on the electric potential difference between iron powder and biochar powder in the SMBs, the ICM can generate abundant tiny galvanic cells with the production of ferrous ions (Fe²⁺) and hydrogen radicals ([H]), which at the anode (ZVI) generate electrons and cathodes (biochar) with the exception of electrons, respectively [34]. These strong oxidizing products could effectively break down the structure of organic molecules [35]. Therefore, the ICM in the SMBs could improve the biodegradability of sewage and greatly enhance the microbial transformation of the pollutants. The pollutants could also be removed through adsorption and from the precipitation by the ferrous and ferric hydroxides formed from oxidation and precipitation [36]. In addition, the ICM could benefit denitrification and NO₃⁻-N reduction, based on the electrochemical reaction and generated electron donors on the surface of the electrodes [30,37]. The possible reaction pathways were shown as follows [28,29,38]:

Anode (Iron):

$$\mathrm{Fe}-2{\mathrm{e}}^{-}\to {\mathrm{Fe}}^{2+}$$

(1)

Cathode (Carbon):

$${\mathrm{O}}_2+4{\mathrm{H}}^{-}+4{\mathrm{e}}^{-}\to 2\mathrm{O}+4\left[\mathrm{H}\right]\to 2{\mathrm{H}}_2\mathrm{O}$$

(2)

$${\mathrm{O}}_2+2{\mathrm{H}}_2\mathrm{O}+4{\mathrm{e}}^{-}\to 4{\mathrm{OH}}^{-}$$

(3)

Consequent reaction:

$${\mathrm{Fe}}^{2+}+2{\mathrm{OH}}^{-}\to \mathrm{Fe}{\left(\mathrm{OH}\right)}_2$$

(4)

$$4\mathrm{Fe}{\left(\mathrm{OH}\right)}_2+2{\mathrm{H}}_2\mathrm{O}+{\mathrm{O}}_2\to 4\mathrm{Fe}{\left(\mathrm{OH}\right)}_3$$

(5)

The variation of the EC values in the MSL systems is shown in Figure 4. The EC value represented the total amount of the soluble salt ions in the sewage [27], which was mainly related to the different types of iron ions generated from the ICM reaction, especially in the MSL systems. During the whole experiment, there were two significantly different trends but totally increased the EC values in the MSL systems. It was attributed to the increasingly efficient chemical processes, especially the electrochemical reaction of the ICM. MSL1 and MSL3 showed higher and steady increases in the EC values with ranges from 2.01 to 2.51 uS/cm and 2.09 to 2.66 uS/cm, respectively. However, the EC variations in MSL2 and MSL4 were lower with ranges from 1.63 to 1.97 uS/cm and 1.69 to 2.08 uS/cm, respectively. In the traditional mechanism analyses, the high DO concentration and high ORP value in the MSL systems were advantageous to the natural oxidation of the zero-valent iron in the humid SMB and the ferrous ions in the sewage [9,12]. However, the experimental results indicated that the bottommost SMB submersion in the MSL system played the more dominant role in transforming ZVI into an ion state. In addition, the submersion could further facilitate the electrochemical oxidation of the intermediate iron ions during the process of the ICM, in this study. Such an interesting result was discovered, and it was mainly due to the water flows around the ICM, rather than the moist conditions in the middle and upper SMBs. Therefore, there were nonnegligible effects with both advantages and disadvantages of the submerged bottommost environment on the effluent quality.

3.3. Pollutant Removal in MSL Systems

3.3.1. COD Removal

As a miniature soil ecosystem, the ICM-based SMB was also the habitat for various microorganisms, which undertook the predominant function in the MSL system for degrading the organic pollutants [14,18]. The removal of COD among the MSL systems is shown in Figure 5a. COD removal in MSL2 and MSL4 increased from the beginning to the end and was overall higher than that in MSL1 and MSL3. In the stable operation phase of MSL2 and MSL4, during day 60 to 80, their average COD removal reached up to 94.2 and 96.0%, respectively, while the initial removal efficiencies of COD were at 88.8 and 90.8%. The novel MSL systems, especially the upper SMBs and PLs, as porous media, had sufficient minute pores and adsorption spaces for trapping organics and microorganisms [11,26]. Though there were slight fluctuations of the COD removal in MSL1 and MSL3 during initial and final experimental period, the COD removal maintained the increasing trends with average efficiencies of 80.9 and 83.1% during day 0 to 30, respectively, as well as 86.1 and 86.0% during day 60 to 80. The microbial activities, such as their growth, domestication, and distribution were gradually strengthened despite the fact that they were invisible [19]. Obviously, the bottommost SMB submersion led to low a DO concentration and reductive environment in the MSL systems and they had a negative impact on the microbial aerobic metabolisms. The oxygenated PLs were advantageous for the organic removal [17]. Moreover, there were two levels of influent COD concentrations, which also led to the differences in the organics removal among the MSL systems, such as MSL1 and MSL3, or MSL2 and MSL4. In sum, the MSL systems had a good performance on the COD removal.

3.3.2. TP Removal

The removal of TP among the MSL systems is shown in Figure 5b. During the whole experiment, the TP removal efficiencies in the MSL systems were maintained at a high level and ranged from 96% to the ideal 100%. It was largely attributed to the promotion of the ICM related to the steady electrochemical reaction without the impacts of the extreme pH and anoxic conditions [19]. The ICM also enhanced the TP removal from all MSL systems, even in the semi-closed systems with the bottommost SMB submersion and the weakly oxidizing environment. Based on the MSL systems coupled with the ICM, the formation of the colloidal Fe (OH)₃ adsorbents and the generation of the final FePO₄ precipitate, could be assured by removing the phosphate radicals with effect [11,37].

3.3.3. Nitrogen Removal

The removal efficiencies of NH₃-N, NO₃⁻-N, and TN among the MSL systems are shown in Figure 6a–c, respectively. It can be significantly observed that the odd-coded MSL systems (MSL1 and MSL3) and the even-coded MSL systems (MSL2 and MSL4) were divided into two different trends for the NH₃-N and NO₃⁻-N removal, which were simply connected with the presence of the bottommost SMB submersion, or by the absence of a bottommost SMB submersion. The oxygenated MSL2 and MSL4 showed the continuous enhancement of nitrification with the range of the NH₃-N removal from 75.4 to 93.8% and 78.5 to 92.3%, respectively. However, the denitrification of the even-coded MSL systems was constrained with ranges from 66.2 and 68.5% to 6.2 and 1.5%, respectively. On the contrary, the NH₃-N removal of the odd-coded MSL1 and MSL3 were progressively decreased from 36.9 and 30.8% to 13.8 and 18.5%, respectively. However, their removal efficiencies of NO₃⁻-N were significantly increased from 66.2 and 68.5% to final 83.8 and 93.1% due to submersion. In the MSL systems, the microbial nitrification and denitrification were the oxidizing process of the NH₃-N and reductive process of NO₃⁻-N, respectively [10,15]. ICM had inapparent influence on NH₃-N removal, which had the significant effect on the NO₃⁻-N removal due to its efficient supply of electron donors for the denitrification [18,26,29]. The slight differences between the two odd-coded or even-coded MSL systems were mainly influenced by the effects of the key operating factors on the ICM.

Denitrification is the key process of nitrogen biotransformation for the TN removal [8]. However, the variation trend of the TN removal was almost the same for the NH₃-N removal due to its proportion of concentration value. Initially, there were rapid processes for the nitrogen removal through the physicochemical adsorption and cation exchange capacity [16,17]. However, during the later stable period, the nitrification and denitrification in the aerobic PLs and anaerobic SMBs were the primary drivers for the biotransformation of NH₃-N and NO₃⁻-N [1,26], respectively. In the experiment, the increased removal efficiencies of NH₃-N and TN in MSL2 and MSL4, the increase of the NO₃⁻-N removal in MSL1 and MSL3 as well, could signal the gradually activated and enhanced microbiological processes to a great extent. Comprehensively, the mechanisms of the pollutants removal in the ICM-based MSL systems could be concluded, as shown in Figure 7.

The variation of and the final C/N ratio of the effluents in the MSL systems is shown in Figure 8, respectively. The average C/N ratios of MSL1 and MSL2 are 1.20 and 1.58, respectively, and were higher than those of MSL3 and MSL4 with values of 0.70 and 0.62, due to the high COD concentration in the sewage. The C/N ratio of the effluents in the MSL systems were totally decreased during the experimental period. During the later stable operation of the experiment, the removal efficiencies of the MSL systems on the organics and the nutrients, were gradually restored on the basis of the environment-adapted microorganisms under a condition with multiple factors. Despite the slight fluctuations, there were still similar trends in MSL3 and MSL4 on the C/N ratio variations during day 70 to 80, with average values of 0.57 and 0.51. However, there were differences in the C/N ratio between MSL1 and MSL2, mainly because of the effect of the ICM on the denitrification and the further TN removal [18]. Finally, the average C/N ratio of the effluent in MSL2 during day 60 to 80 was 1.32, which was higher than that of effluent in MSL3, which was 1.00. In consideration of the final effluent quality, the C/N ratios ranged from 0.46 to 1.52, other land-based and eco-technologies, such as CW can be applied and followed closely to the MSL systems for further NO₃⁻-N removal [13,15].

3.4. Effects of the Operating Factors on the ICM-Related Water Parameters and Pollutants Removal among the MSL Systems

3.4.1. Main Effect

In the factorial analyses, the statistical significance of the main effects and the interactive effects can be defined [25]. As shown in Figure 9, for eight responses of concern in the normal plots of residuals for the factorial analysis, the points fell approximately along a straight line, which indicated the normal distribution and a reliable set of experimental data for further factorial analysis [39]. Additionally, the ANOVA results for the factorial analyses of the responses that pH, ORP, EC, and the removal efficiencies of COD, TP, NH₃-N, NO₃⁻-N, and TN, are shown in

Table 2. Analysis of variance (ANOVA) and the standardized effect of factors for the factorial analysis on eight pollutants removal (pH, ORP, EC, and the removal efficiencies of COD, TP, NH₃-N, NO₃⁻-N, and TN).

Response	R1: pH
Source	F-Value	p-Value Prob > F	Std Contribution (SC1; %)
Factor-A	306.59	<0.0001	45.90
Factor-B	230.90	<0.0001	34.56
Interaction of AB	122.53	<0.0001	18.34
Response	R2: ORP
Source	F-Value	p-Value Prob > F	Std Contribution (SC2; %)
Factor-A	84.19	<0.0001	8.24
Factor-B	869.80	<0.0001	−85.17
Interaction of AB	59.30	<0.0001	−5.81
Response	R3: EC
Source	F-Value	p-Value Prob > F	Std Contribution (SC3; %)
Factor-A	3.09	0.1170	−3.05
Factor-B	90.00	<0.0001	88.92
Interaction of AB	0.12	0.7344	0.12
Response	R4: removal efficiency of COD
Source	F-Value	p-Value Prob > F	Std Contribution (SC4; %)
Factor-A	2.41	0.1589	−0.81
Factor-B	283.56	<0.0001	−95.40
Interaction of AB	3.25	0.1092	1.09
Response	R5: removal efficiency of TP
Source	F-Value	p-Value Prob > F	Std Contribution (SC5; %)
Factor-A	0.14	0.7153	−0.49
Factor-B	17.29	0.0032	−59.61
Interaction of AB	3.57	0.0955	12.32
Response	R6: removal efficiency of NH3-N
Source	F-Value	p-Value Prob > F	Std Contribution (SC6; %)
Factor-A	1.641 × 10−13	1.0000	1.121 × 10−14
Factor-B	1455.53	<0.0001	−99.45
Interaction of AB	0.068	0.8013	4.623 × 10−3
Response	R7: removal efficiency of NO3−-N
Source	F-Value	p-Value Prob > F	Std Contribution (SC7; %)
Factor-A	0.28	0.6106	0.063
Factor-B	434.06	<0.0001	97.77
Interaction of AB	1.60	0.2416	−0.36
Response	R8: removal efficiency of TN
Source	F-Value	p-Value Prob > F	Std Contribution (SC8; %)
Factor-A	0.011	0.9179	6.439 × 10−4
Factor-B	1748.21	<0.0001	−99.45
Interaction of AB	1.59	0.2427	−0.091

, corresponding to R₁, R₂, R₃, R₄, R₅, R₆, R₇, and R₈, respectively. The statistical “p-values” of the factorial models for R₁–R₈ were <0.0001, <0.0001, <0.0001, <0.0001, 0.0126, <0.0001, <0.0001, and <0.0001, respectively. The lower a “p-value” of 0.05, the more significant the factorial model for the responses [40]. The absolute value of the std contribution represents the magnitude of the factorial effect on the response [8]. The plus and minus preceding the values of the standardized (std) contributions are indicative of the positive and negative effects of the factor on the response, respectively [19].

The normal plots of the factorial effects on eight responses for the MSL systems are shown in Figure 10. The point on the left or right sides of the dividing line indicated their featured negative (blue) or positive (orange) effects, respectively [14]. The vertical distance from the point to the dividing line represented its significant levels, and there was almost no significance of the effect when the point was on the dividing line [41]. These results were highly consistent with the main effect on the responses in Figure 11. In detail, Factor A-influent: the C/N ratio had the most significant positive effect (p < 0.0001, SC₁ = 45.90%) on pH (R₁). The higher value of the C/N ratio, in other words, the more plentiful the carbon source for denitrification, the relatively faster consumption of the acidity in the sewage [8]. Similarly, Factor B-bottommost SMB submersion: was the positive effect (p < 0.0001, SC₁ = 34.56%) on R₁ due to its high level with the anaerobic and reductive conditions, as well as the efficient ICM in the SMB could promote the denitrification process [10,14]. Factor A also had the most significant positive effect (p < 0.0001, SC₂ = 8.24%) on ORP (R₂), while Factor B had the most significant negative effect (p < 0.0001, SC₂ = −85.17%) on R₂. ORP was closely related to the redox environment in the MSL system, the high level of Factor A and Factor B led the dominant oxidizing and reductive conditions, respectively. Contrary to ORP, factor B had the most significant positive effect (p < 0.001, SC₃ = 88.92%) on EC (R₃), however, Factor A was a non-significant effect (p = 0.1170, SC₃ = −3.05%). Based on the mechanisms of the ICM, the value of EC was increased along with the production reinforcement of iron ions.

As shown in Figure 10, Factor B had the most dominant and significant negative effects on the removal efficiencies of COD (R₄) (p < 0.0001, SC₄ = −95.40%), TP (R₅) (p = 0.0032, SC₅ = −59.61%), NH₃-N (R₆) (p < 0.0001, SC₆ = −99.45%), and TN (R₈) (p < 0.0001, SC₈ = −99.45%), respectively. The high height of the effluent outlet and the high-level submersion resulted in relatively low DO and ORP within the weakly reductive MSL systems, which were disadvantageous for the aerobic microbial or physicochemical processes. In particular, Factor B had the significant positive effect on the removal efficiency of NO₃⁻-N (R₇) (p < 0.0001, SC₇ = 97.77%). In addition, there were non-significant effects of Factor A on all pollutants’ removal (R₄–R₈). For conventional soil-based technologies, the influent C/N ratio played the key role in adjusting the denitrifying efficiencies of the functional microbes [8,9]. However, the nitrogen removal process was almost undisturbed under the condition of the slight differences of the influent C/N ratio. It was mainly attributed to the PBS as the novel carbon source in the anaerobic SMBs and the effective ICM for consistently providing electron donors with denitrifying bacteria. It also indicated that the microelectrolysis-based SMBs in the MSL systems can ensure the stable and efficient performance on sewage treatment.

3.4.2. Interactive Effect

Based on the factorial analysis [18,41], the numeric relation between the factors and the responses can be simulated and shown as following a polynomial regression Formula (6). The factorial polynomial regression equations for eight responses are shown as following (7) to (14) with the R-squared (R²) of the factorial equations of 0.9880, 0.9922, 0.9210, 0.9731, 0.7241, 0.9945, 0.9820, and 0.9898, respectively.

$$R_e or V=b_0+b_{1}A+b_{2}B+b_{3}AB$$

(6)

$$V\left(pH\right)=7.55+0.29A+0.25B+0.18AB$$

(7)

$$V\left(ORP\right)=67.50+9.33A-30.00B-7.83AB$$

(8)

$$V\left(EC\right)=2.25-0.042A+0.23B+8.333E\left(-0.03\right)AB$$

(9)

$$R_e\left(COD\right)=90.58-0.42A-4.52B+0.48AB$$

(10)

$$R_e\left(TP\right)=97.17-0.17A-1.83B+0.83AB$$

(11)

$$R_e\left(N{H}_3-N\right)=55.38+0.000A-36.67B-0.25AB$$

(12)

$$R_e\left(N{O}_3^{-}-N\right)=50.50+1.03A+40.63B-2.47AB$$

(13)

$$R_e\left(TN\right)=53.49+0.058A-22.93B-0.69AB$$

(14)

Figure 12 shows the interactive effects between Factors A and B on eight responses. There may be no interactions when two lines approximately in parallel [25], such as the effect of the interaction AB on R₃–R₈, which also confirm the results in Table 2 and Figure 10. However, there was the significant positive and negative effects of the interaction AB on pH (R₁) (p < 0.0001, SC₁ = 18.34%) and ORP (R₂) (p < 0.0001, SC₂ = −5.81%), respectively, attributed to the potential point of the intersection on the extension of two lines with different slopes. Instead of the simple superposition of the main effect, there was comprehensive synergy or antagonism included in the factorial interaction [19]. It was suggested that the effect of the significant main factor was larger than the factorial interaction on the pollutants removal in the MSL systems. In addition, the significant interaction was mainly related to the externally conditional parameters such as pH and ORP other than the pollution index. Therefore, the scientific mechanisms for the actual removal of the pollutants in the sewage should be paid more special attention, based on the statistical analysis results of the experimental data. The different levels of the operating factors applied in the MSL systems would drive the trends in the performance variation of the systems comprehensively. Thus, it is necessary to select more operating factors that can directly improve the efficiency of the microbial and physicochemical mechanisms. In terms of the performance optimization of the ICM-based MSL system, it is also important to adjust the relevant factors to their optimized level through a multiple factorial design.

4. Conclusions and Future Perspectives

The ICM-based MSL system is characterized by 5E (economical, efficient, eco-friendly, easy-to-construct, and energy-saving), which is a suitable and sustainable option for the decentralized domestic sewage treatment in remote and small communities in developing countries. This study firstly introduced microelectrolysis into explaining the internal mechanisms of pollutant removal in the MSL system. The result showed that the higher influent C/N ratio, the larger increase of pH. Furthermore, the ORP value could be more intuitive to reflect the redox properties in the MSL system than the DO concentration. It was proved that the unimpeded water flow and a high ORP value were advantageous to the microbial metabolic degradation of organics. The increase of EC in the MSL systems was attributed to the efficient electrochemical reaction of the ICM. The ideal TP removal was largely attributed to the promotion of the ICM with a steady electrochemical reaction, even in the semi-closed systems with the weakly oxidizing environment. The ICM had a non-apparent influence on the NH₃-N removal, while it had a significant effect on the NO₃⁻-N removal due to its efficient supply of electron donors. The nitrogen removal process was almost undisturbed under the condition of the slight differences of the influent C/N ratio. Denitrification was simply connected with the presence of the bottommost SMB submersion, or by the absence of a bottommost SMB submersion. The C/N ratio of the effluents in the MSL systems were wholly decreased during the experimental period. The results of the factorial analyses indicated that the significant interaction was mainly related to the externally conditional parameters such as pH and ORP. The effects of the significant main factor was larger than the factorial interaction on the pollutants removal in the MSL systems.

In this study, the robust performance of the well-operated MSL system, with respect to sewage treatment under multivariate operating factors, was attributed to its innovative brick-pattern structure, its improved dimensional design, clear division of the functional areas for the microbial and physicochemical processes, and the advantageous microelectrolysis mechanisms. The results of the factorial analysis provided a theoretical foundation for the optimal design and operation of on-site MSL systems. What’s more, the MSL system is also appropriate for the new-style leisure eco-tourism bases throughout rural China and can minimize the pollution on the local natural water environment, the treated water could be also reused for planting and landscaping. We can look at the efficient treatment of rural sewage containing PPCPs such as antibiotics and surfactants may be also achievable through the ICM-based MSL system in the near future.