Effect of hydraulic conditions on manganese accumulation by physical and chemical pathways in chlorinated water distribution system: A preliminary laboratory study

‘Black water’ is a common discoloration problem of drinking water (Sly et al. 1990; Cerrato et al. 2006; Li et al. 2019), which is caused by the resuspension of accumulated manganese (Mn) on the inner surfaces of drinking water distribution systems (DWDS; Kohl & Medlar 2006). Significant accumulation of Mn has been observed in operational DWDS even when Mn concentration in treated water was as low as approximately 10 μg·L⁻¹ (Li et al. 2019; Ma et al. 2019). This is much lower than the typical guideline values of 50 μg·L⁻¹ or 100 μg·L⁻¹ worldwide, which complicates controlling Mn accumulation and subsequent discoloration problems.

Because the chemical pathway involves the oxidation of Mn²⁺, Mn accumulation processes could be affected by water characteristics, e.g., pH and coexistent substances (Hao et al. 1991; Allard et al. 2013; Li et al. 2020), as well as Mn concentration and speciation (Hao et al. 1991; Li et al. 2019; Zhou et al. 2020).

Hydraulic conditions in distribution pipes have been suggested as an important factor in material accumulation in DWDS (Vreeburg & Boxall 2007; van Thienen et al. 2011). High water velocity in pipe networks has been claimed to be effective in controlling accumulation of suspended solids by creating high shear stress (Boxall & Saul 2005). Vreeburg (2007) proposed a ‘self-cleaning pipe network’, in which material accumulation was prevented by achieving the daily maximum water velocity of 0.4 m·s⁻¹. Nonetheless, the influence of hydraulic conditions on Mn accumulation has rarely been investigated, except in the study by Sly et al. (1988), who examined the effect of water velocity on the development of the Mn-depositing biofilm in DWDS. Furthermore, the effect of hydraulic conditions on the physico-chemical parameters of Mn accumulation, specifically via the aforementioned physical and chemical pathways, has not been studied yet.

The objective of this study was to understand the influence of hydraulic conditions on Mn accumulation through physical, chemical pathways, and Mn detachment. Apart from our previous study (Zhou et al. 2020), which tested Mn accumulation in a clean batch system, we set up a semi-continuous system fed with real drinking water to imitate its flowing condition. Kinetics of these processes were then described by a numerical model based on our previous study (Zhou et al. 2020), and the effect of hydraulic conditions was formulated. The obtained results underscored the significance of the influence of hydraulic conditions on Mn accumulation in well chlorinated DWDS and helped to suggest appropriate countermeasures for preventing Mn accumulation.

Manganese sulfate (MnSO₄), sodium hypochlorite (NaClO), and nitric acid (HNO₃) used in this study were of reagent grade and purchased from FUJIFILM Wako Pure Chemical (Osaka, Japan). A stock solution of MnSO₄ (0.5 g-Mn²⁺·L⁻¹) for Mn accumulation experiments was prepared with ultrapure Milli-Q water using a Milli-Q Reference Water Purification System (Merck Millipore; Burlington, MA) and stored at 4 °C in the dark. Pure water (Elix water) used in the detachment experiment was produced by an Elix Advantage Water Purification System (Merck Millipore, Burlington, MA). The tap water used in this study was obtained from a laboratory at Kyoto University.

A laboratory-scale reactor was used to investigate the influence of water hydraulic parameters on Mn accumulation and detachment (Fig. S1 in the Supplementary Materials). It was a vertical acrylic column (length, 70 cm; inner diameter, 2.5 cm), containing a plastic (polyvinyl chloride) holder in the center that held test pieces (1.5 cm in width) on both sides, which minimized the influence of gravity, a less important contributor to the accumulation of substances in DWDS (Boxall & Saul 2005; Vreeburg 2007). Water was fed from the bottom to the top of the reactor. During the experiments, the reactor was covered to prevent the influence of light. All the experiments were performed at room temperature (20 °C).

Test pieces made of epoxy resin, kindly provided by KUBOTA Corporation (Osaka, Japan), were used in both the accumulation and detachment experiments (see the following sections). This material was selected because it is one of the most common lining materials for ductile iron pipes in Japan (JWWA 2012). Two types of test pieces with different surface areas were used. Eight test pieces with a 7.5 cm² (1.5 cm × 5 cm) and 16 test pieces with 22.5 cm² (1.5 cm × 15 cm) were evenly inserted on both sides of the plastic holder in the detachment and accumulation experiments, respectively. Vertical difference in Mn accumulation within the reactor was considered negligible because the formed particles were so small that sedimentation would be inhibited by upward water flow. This was partially confirmed during preparation of the detachment experiment (see below).

The reactors and other system components (e.g., tanks and pumps) were soaked in 0.1 M HNO₃ overnight and then rinsed with tap water before use. The test pieces were sonicated for 15 min and then rinsed with Milli-Q water to prevent the introduction of extra particles.

The accumulation experiments were performed at four velocities, 0.02, 0.05, 0.08, and 0.20 m·s⁻¹. Reynolds number was estimated from the hydraulic diameter of reactor (as calculated considering the plastic folder inside), kinematic viscosity of water at room temperature (20 °C), and flow velocity. These velocities corresponded to Reynolds numbers ranging from 230 to 2,300, which indicated that the reactors were in laminar to transient flow regimes. Although most of the real distribution pipes are considered to be in the turbulent flow regime, we believe that our experimental conditions were still useful for understanding the qualitative effects of flow characteristics on Mn accumulation. The entire study was divided into two groups of experiments. Mn accumulation at water velocities of 0.05 and 0.2 m·s⁻¹ was examined in Group 1 experiments, whereas accumulation at velocities 0.02 and 0.08 m·s⁻¹ was examined in Group 2 experiments. All experiments lasted for 4 weeks.

The detachment experiment was conducted in two steps: the formation and detachment of accumulated Mn. The same reactors were used in both steps. Figure S2 shows the schematic of the detachment experiment.

Accumulation of Mn that would subsequently detach proceeded under the flowing conditions of tap water containing Mn²⁺ and free chlorine (Fig. S2a). MnSO₄ and HOCl were dissolved and mixed in 50 L of tap water in a polyethylene tank with final concentrations of 1 mg-Mn²⁺·L⁻¹ and 6 mg-Cl₂·L⁻¹, respectively. High concentrations of Mn²⁺ and Cl₂ were used to rapidly accumulate Mn on the test pieces. The Mn solution was continuously pumped into four parallel reactors at a water velocity of 0.02 m·s⁻¹, and the effluent of the reactors was recirculated to the polyethylene tank (Fig. S2a). To maintain constant water parameters, such as pH, Mn concentration, Mn speciation, and Cl₂ concentration during the accumulation process, the feed water was freshly prepared every day. The entire process lasted for 7 days. Subsequently, test pieces at the top, middle, and bottom of each reactor were collected, and average value of the three positions was calculated to determine the initial amount of Mn in each reactor before detachment. There were no significant differences among the three positions in the amount of accumulated Mn in all reactors.

The plastic holders with test pieces as prepared above were immediately transferred to other clean columns, as shown in Figure S2b. To observe the detachment process selectively, Elix water was used as feed water in this step, as it was expected to contain negligible concentrations of particulate matter and Mn to cause the accumulation process. Elix water (50 L) was pumped into four reactors from a plastic tank at different velocities (0.02, 0.05, 0.15, and 0.30 m·s⁻¹). The effluent of the reactors was recirculated into the plastic tank. To prevent the re-attachment of detached Mn from the test pieces, Elix water was exchanged every day in the first week and then every other day. Test pieces were collected in triplicate on days 3, 10, 20, and 30 to observe the time course of the detachment of accumulated Mn.

Mn that accumulated on the test pieces in all experiments was recovered by soaking each test piece in 90 mL of 1 M HNO₃ for 7 days. The amount of accumulated Mn was calculated from its concentration in the solution. The aqueous Mn in the reaction tank in the accumulation experiments was fractionated into total Mn and soluble Mn. The latter was defined as the fraction less than 0.2 μm in diameter (GL Chromatodisc 13P; GL Science, Tokyo, Japan), whereas the difference between total and soluble Mn (Mn²⁺) reflected the insoluble Mn (MnO₂). Fractionation was immediately conducted after collecting the water sample to prevent further chlorination of Mn²⁺. In Group 2, we also investigated Mn samples passed through a 10 kDa membrane (Amicon^® Ultra 15 mL Centrifugal Filters – 10 kDa, Merck Millipore) to confirm the solubility of Mn in particles below 0.2 μm in diameter. All the samples for Mn analysis were digested with 0.5 M HNO₃ at 105 °C for 7 h (American Public Health Association (APHA) 2005) and then stored at 4 °C in the dark before analysis. The concentrations of Mn in the samples were measured by inductively coupled plasma mass spectrometry (Xseries 2Xt, Thermo Fisher Scientific; Waltham, MA). The pH was measured using a pH meter (pH METER D-51, Horiba, Kyoto, Japan). Residual chlorine was analyzed by the N,N-diethyl-p-phenylenediamine colorimetric method using a residual Cl₂ meter (AQ-201, SIBATA; Saitama, Japan).

Differences between the initial and residual amounts of accumulated Mn in the detachment experiment and between the concentrations of Mn in particles below 0.2 μm in diameter and in solutions passed through the 10 kDa membrane in Group 2 of the Mn accumulation experiments were analyzed using the paired Student's t-test. The parametric one-way analysis of variance and Bonferroni P-value corrected t-test were performed using Python package statsmodels (https://www.statsmodels.org/) (Seabold & Perktold 2010) to test the difference between fractions of detached Mn after 30-day detachment experiments under each examined water velocity. Differences were considered statistically significant when P < 0.05.

The values of K_det,K_att, and c were independently estimated for each water velocity by fitting the equations to the experimental results. First, K_det was estimated by fitting the integrated form of Equation (8) to the results of the detachment experiment, and the relationship between K_det and water velocity was formulated. Incorporating this relationship into Equations (3)–(8), the values of K_att and c were estimated based on the accumulation experiment results. The week 1 results of the accumulation experiments were used for estimating the K_att value, because the initial contribution of the chemical pathway was considered negligible (Zhou et al. 2020). Finally, the value of c was obtained by fitting the entire set of the accumulation experiment results. The details of the fitting procedures are shown in Supplementary Text S2.

The water characteristics in the reaction tank were monitored daily or every other day. pH was stable in both groups, being 7.5 ± 0.12 (error: standard deviation; the same below) in Group 1 and 7.7 ± 0.24 in Group 2. Residual Cl₂ became stable at 0.62 ± 0.15 mg-Cl₂·L⁻¹ and 0.9 ± 0.3 mg-Cl₂·L⁻¹ (before day 22) in Groups 1 and 2, respectively (Fig. S7). In contrast, because of Mn accumulation in the whole system (i.e., reactor walls, tubing, and pumps), the total Mn concentration in the reaction tank decreased in the first week from 6.0 and 10 μg·L⁻¹ to 0.73 and 1.6 μg·L⁻¹ in Groups 1 and 2, respectively, and then stabilized at these low concentrations (Fig. S8). The sudden increase in the concentrations of total Mn and residual Cl₂ in Group 2 was due to the temporal stoppage of pumps for system maintenance. The concentration of Mn particles below 0.2 μm in diameter also decreased in the first week and then stabilized at 0.57 ± 0.16 and 0.80 ± 0.66 μg·L⁻¹ in Groups 1 and 2, respectively (Fig. S8). Mn concentrations in the fraction that passed through 0.2 μm pore size and those that passed through 10 kDa membrane in Group 2 were not significantly different (P > 0.05, Fig. S9), suggesting that most Mn smaller than 0.2 μm pore size were also smaller than 10 kDa. As it is technically difficult to discriminate ionic and colloidal Mn in such a small size range, the fraction of Mn smaller than 0.2 μm pore size was considered as Mn²⁺ in this study.

In the range of the examined flow conditions, the c value ranged from 5.6 to 10 M^−1.3 s⁻¹, which was one order of magnitude smaller than that that obtained in preliminary experiments using conical flasks, 1.0 × 10² M^−1.3 s⁻¹ (Table S2). In contrast, K_att ranged from 9.1 × 10⁻⁸ to 6.2 × 10⁻⁷ cm⁻² Ls⁻¹, which was larger than that in preliminary experiments, 1.6 × 10⁻⁸ cm⁻² Ls⁻¹ (Table S2). The comparison suggested that different experimental conditions significantly impacted the two separate pathways of accumulation.

In this study, the effects of hydraulic conditions on the physical and chemical pathways of Mn accumulation were investigated for the first time. In the accumulation experiments, the characteristics of water flowing into the reactors were controlled so that both Mn²⁺ and MnO₂ were present. Consequently, Mn accumulation increased linearly with time approximately during the first 7 days of operation, and after that time point, it accelerated, especially at the water velocity of 0.2 m·s⁻¹ (Figure 4). Although such acceleration is often explained by the biological accumulation through Mn-oxidizing bacteria (Sly et al. 1990; Li et al. 2022), biological processes most likely were negligible in our experiments because the residual chlorine was maintained at a high concentration (0.6 − 0.8 mg·L⁻¹) (Fig. S7). Therefore, these observations strongly indicate the presence of two pathways of Mn accumulation: the physical pathway, in which MnO₂ particles attached to the surface linearly with respect to time, and the chemical pathway, i.e., the autocatalytic oxidation of Mn²⁺ on the surface of accumulated Mn (Equations (1) and (2)). The presence of these two pathways was also observed in our previous study (Zhou et al. 2020).

As mentioned in the Results section, the estimated values for K_att and c in the accumulation experiments using reactors (Table S3) were 5.7−39 times larger and 10−18 times smaller than those obtained in preliminary experiments using conical flasks (Table S2), respectively. The dissimilarity of K_att values can be attributed to the differences in the surface-to-volume ratios, which are higher in the reactor than in the conical flasks. The higher surface-to-volume ratio of the reactor might have led to enhanced transportation of MnO₂ particles via the physical pathway.

In contrast, the lower rates of the chemical pathway in the reactors were presumably caused by the differences in water matrix: tap water and carbonate buffer (0.5 mM NaHCO₃) were used in the reactor and conical flasks, respectively. The hardness of tap water was approximately 40 mg·L⁻¹ (Kyoto Water Supply and Sewage Bureau 2020), which suggests the presence of Ca²⁺ and Mg²⁺ cations. Huangfu et al. (2013) reported that even low concentrations of Ca²⁺ and Mg²⁺ (i.e., <1 mM) were significantly more efficient than 10 mM Na⁺ in enhancing the aggregation of MnO₂ particles. Thus, larger MnO₂ particles might have formed in the tap water than in the carbonate buffer in the preliminary experiments (Text S1), which could lead to the deceleration of autocatalytic oxidation because of the lower specific surface area of the formed MnO₂.

The flow in the accumulation experiments was mostly laminar, except in the fastest, 0.2 m·s⁻¹ condition, which could be considered as transient to turbulent flow. Although operational DWDSs are generally considered to be within the turbulent flow regime, we selected these low velocity conditions as it was difficult to control the speciation of Mn (the proportion of Mn²⁺ to total Mn) in the feed water under high turbulence in the current experimental setup. Thus, our results below are limited to non-turbulent conditions. However, such conditions are still possible in the distal areas that have lower water velocities and hence relevant.

According to the experimental results, the higher water velocity promoted both the detachment (Figure 2, Fig. S5) and the total accumulated Mn (Figure 4). The rate constants and related coefficients (c, K_att, and K_det) also positively correlated with water velocity and Reynolds number in the examined ranges (Equations (10)–(12), Figures 3 and 6). These observations indicate that flow conditions strongly influenced the physical and chemical pathways of Mn accumulation and detachment of the accumulated Mn.

Detachment is likely to be caused by the sheer force near the surface of the pipes, which is formed by the velocity gradient perpendicular to the surface, and it becomes more pronounced with an increased water flow velocity in the pipe. Thus, the higher water velocity in the detachment experiments enhanced the detachment process. However, despite this relationship, the total amount of accumulated Mn was still higher under the higher water velocity conditions, which means that the enhancement of the physical and chemical pathways of Mn accumulation was more dominant than that of the detachment process.

As for the physical pathway (i.e., the attachment of MnO₂ to the surface), a particularly high K_att value was estimated at the water velocity of 0.2 m·s⁻¹ compared to those at the lower velocities (0.02, 0.05, and 0.08 m·s⁻¹) conditions (Figure 6). Considering that the flow regime at 0.2 m·s⁻¹ condition was transient flow, turbulent diffusion of MnO₂ particles to the surface of the test pieces was likely to be more significant than at lower velocities and laminar flow conditions. A similar trend was also reported for initial bacterial adhesion to plastic materials (Saur et al. 2017).

The chemical pathway (i.e., autocatalytic oxidation of Mn²⁺ on the pipe surface) is considered to involve two steps: the adsorption of Mn²⁺ to accumulated MnO₂ (i.e., the formation of MnO₂·Mn²⁺) and subsequent oxidation of MnO₂·Mn²⁺ (Equations (1) and (2)). The rates of both processes are extremely rapid (Hao et al. 1991; Van Benschoten 1993), which means that the transport of Mn²⁺ and HOCl to the surface of accumulated MnO₂ is the rate-limiting step in the chemical pathway. Therefore, as in the physical pathway, the enhanced transport of chemical species triggered by higher turbulence might be the main reason for the positive impact of water velocity on the chemical pathway.

The predictions show that Mn accumulation via the chemical pathway continued to accelerate during the 1-month accumulation experiments. However, during the longer-term operation, it can be expected that the surface of accumulated MnO₂ available for Mn²⁺ adsorption would not increase anymore after MnO₂ completely covered the surface of the test pieces. Therefore, this acceleration would be limited by the increasing amount of accumulated Mn (Hao et al. 1991; Zhou et al. 2020), and the entire Mn accumulation process would reach a steady state during the longer-term operation.

Furthermore, during the initial period of accumulation (until approximately day 100), the large water velocity enhanced the accumulation process regardless of Mn speciation because the higher water velocity promoted accumulation via both the physical and chemical pathways, whereas the initial contribution of detachment was lower (Figure 7). On the other hand, the effect of water velocity at a steady state was opposite between two cases of Mn speciation (Figure 8): positive or negative effects were observed under the lower and higher Mn²⁺ concentrations, respectively. This is due to the differences in slopes of the linear relationships between water velocity and the rates of physical and chemical pathways (Figure 6), in which the impact of velocity was larger on K_att than on the c value. When Mn²⁺ proportion was low (Figure 8(a)), the positive impact of velocity on the physical pathway was stronger than on the detachment. On the other hand, in the case of high proportions of Mn²⁺ (Figure 8(b)), the positive impact of velocity on the chemical pathway was lower than on the detachment, leading to the tendency opposite to that seen in the case of low relative Mn²⁺ level. Further predictions at steady state (Fig. S10) showed that the effect of water velocity was dependent on water characteristics (i.e., pH, HOCl concentration, and Mn speciation). In general, the positive influence of water velocity was expected when the water characteristics were less suitable for the chemical pathway to take place, i.e., lower pH, HOCl concentration, and Mn²⁺ proportion. To summarize, these predictions showed that high water velocity could enhance the Mn accumulation process, especially at the initial phase, whereas the subsequent effect of high water velocity would vary depending on the related water parameters.

All the above laboratory-scale results, particularly those that higher water velocity caused larger accumulation, should be interpreted with caution to discuss Mn accumulation in full-scale DWDS. The net effect of water velocity on Mn accumulation in actual pipes would depend on the balance of two positive impacts of water velocity, i.e., those on each accumulation pathway and those on the detachment. Therefore, further research is needed to estimate these effects in actual operational DWDS quantitatively. Nonetheless, it is still very likely that the water flow regime in DWDS can greatly affect the physical and chemical pathways of Mn accumulation. These possibilities are consistent with the previous field observation by Ma et al. (2019), in which they analyzed the deposits in distribution pipes at various distances from a water treatment plant. They observed that Mn accumulation was significantly higher at the sites closer to the plant and lower at the sites further away from the plant, which differed from the profiles observed for other elements (Al, Si, and Fe). Although the authors speculated that the higher residual chlorine and soluble Mn concentration in upstream pipes contributed to the faster chemical oxidation and deposition of Mn, the enhancement of the physical and chemical pathways by higher water velocities in upstream pipes can also be considered as a reason, considering that water velocity is generally higher in the upstream pipes (as also shown by the authors). This preferential accumulation of Mn in the upstream pipes was also reported by Li et al. (2018). Therefore, our results suggest that just increasing the daily water velocity in DWDS may not be a very effective countermeasure for controlling the Mn accumulation, although it is a common strategy for reducing the suspended solids deposition (Vreeburg 2007). Our predictions also indicated that the less suitable water characteristics for the chemical pathway would lead to a lower amount of accumulated Mn (Figure 8, Fig. S10). Thus, adjusting the Mn speciation and water parameters in the treated water could be another option to control Mn accumulation.

We investigated the effect of water hydraulic conditions on Mn accumulation, assuming an ideal situation. It should be noted, however, that several important aspects of actual DWDS were not considered. First, the effect of water velocity and related phenomena might be quite different in case of the low residual chlorine concentration because microorganism-mediated accumulation would become dominant (Li et al. 2022). It also has been suggested that the detachment of accumulated matters is affected by daily fluctuations of hydraulic conditions (Boxall & Saul 2005), which was also not considered in this study. These aspects should be investigated in future research to achieve better understanding of the effect of water hydraulics on Mn accumulation.

In this study, we investigated the influence of water hydraulic conditions on Mn accumulation process in well chlorinated distribution systems. The major findings are summarized as follows:

• – Experiments and modeling analysis confirmed the existence of the physical and chemical pathways for Mn accumulation. In particular, the contribution of the chemical pathway was found to increase with time, leading to accelerated accumulation.

• – The rates of the physical and chemical pathways and that of detachment positively correlated with water velocity in the current experimental setup, suggesting that the hydraulic conditions of the DWDS would play an important role in Mn accumulation and detachment.

• – Long-term prediction of the accumulation experiments indicated that large water velocities promote Mn accumulation, especially in the early stages, but that subsequent effects were variable, depending on the associated water characteristics. It was noted that maintaining high water velocities in DWDS, the common countermeasure against the suspended solids accumulation, might not be appropriate for controlling Mn accumulation.

• – It was also predicted that adjusting the Mn speciation in the treated water, i.e., reducing the proportion of Mn²⁺ to total Mn, would effectively control the Mn accumulation in DWDS, as also suggested in our previous study (Zhou et al. 2020).

This work was partially supported by grant 18H05918 from the Japan Society for the Promotion of Science (JSPS) KAKENHI. The authors would like to thank KUBOTA Corporation for kindly providing the test pieces of the pipe materials.

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

The authors declare there is no conflict.