Effects of highly corrosive pipe wall and disinfectant constituent on the chlorine decay behavior in drinking water

The drinking water quality would get better than source water after a series of treatment in the drinking water treatment plant (DWTP). The treated water would be delivered to the thousands of households by resorting to the DWDS. The long-used pipelines could serve as biochemical reactor and make water quality deterioration to different extents, which depends on pipe materials and service age. This phenomenon is particularly significant for the stagnant water points or end points within the distribution system. Even active disinfection is widely taken by dosing chlorine-containing disinfectant, and the pipe wall-induced water quality deterioration could not be fundamentally avoided.

Chlorine-containing disinfectant could kill or inactivate microorganisms, so their residual level has become the key water quality index concerned by relevant regulators. It is reported that total chlorine residual (TCR) decay behavior may be remarkably different for the commonly used pipe materials. Digiano & Zhang (2005) developed a bench-scale pipe section reactor and found that free chlorine decay in the cast iron pipe is faster than that in the cement-lined ductile iron pipe. What is more, the model calculation results showed that wall-induced TCR decay accounts for a larger proportion relative to the loss in the bulk phase. With the help of the same reactor, the cast iron pipe was found to force chloramine to decay faster than cement-lined ductile iron pipe, and the chloramine decay was three orders of magnitude faster than bulk decay in the presence of cast iron pipe (Westbrook & Digiano 2009). A water quality model was developed in EPANET to model the full-scale DWDS in the town of Goderich, and it was found that the largest impacts are from the unlined cast iron pipe and lesser impacts are progressively from the ductile pipe and then polyvinylchloride pipe (Huang & McBean 2008). Another field study on chloraminated distribution systems further corroborated that unlined cast iron pipe consumes chloramine more severely than epoxy-lined cast iron pipe and PVC pipe (Liu et al. 2015). Therefore, aged cast iron pipe serves as the main cause to the disinfectant decay in the full-scale water distribution system.

The cast iron pipe is characterized by a rough inner wall consisting of iron tuberculation and attached biofilms (Li et al. 2014). On encountering fresh bacteria, chlorine/chloramine could oxidize cellular membranes, nucleic acid, and other living matters until it is killed, which is the fundamental purpose of using disinfectant in the field of DWDS. Besides, the microorganism within the biofilm can also produce, release, and accumulate extracellular polymeric substances to protect microbes from disinfectants and toxic materials as a response to chloramine stress (Herath & Sathasivan 2020). Hence, the existence of biofilm attached to the pipe wall would necessarily consume disinfectant to some extent depending on the total biomass and metabolic activity. On the other hand, the exposed zero-valent iron can reduce chlorine/chloramine to Cl⁻ in the way of an electrochemical reaction, which appears to be a disinfectant-consuming process as well (Li et al. 2014). The complicated surface composition makes the cast iron pipe wall act as a ‘black hole’ to intensively consume free chlorine or combined chlorine. In fact, the surface composition could be reshaped in response to the disinfectant constituent and concentration. It is reported that chloramine disinfection is more prone to stimulate nitrifiers to grow on the pipe wall (Hua et al. 2011) and produce dense crystallized particles (Li et al. 2014) relative to chlorine disinfection. Once established in biofilms, nitrifiers can be difficult to eliminate by solely increasing chloramine concentrations, as the nitrifiers are relatively resistant to the dissolved chloramine by releasing soluble microbial products (Krishna et al. 2012). However, free chlorine could stop biofilm development in organic carbon-limited water (Chandy & Angles 2001). Considering the adaptability of biofilm to the specific disinfectant (Shen et al. 2016), periodic free chlorine maintenance has been suggested and applied in chloraminated water supply systems to stop nitrification (Hua et al. 2011). The reaction of surface composition to disinfectant alteration should be further defined for the aged CI pipe to put forward applicable water treatment and system management strategies.

Actually, many utilities have now switched from chlorine gas and liquid ammonia to sodium hypochlorite (NaClO) and ammonium sulfate ((NH₄)₂SO₄), respectively, due to security concerns. The impact of transition in disinfectant constituent on pipe wall composition and further on the disinfectant decay behavior is still in the course of clarification. To investigate the effects of aged CI pipe wall and disinfectant constituent on the chlorine decay behavior, the bulk decay behavior was studied using brown glass bottles for NaClO, NaClO + NH₃·H₂O, and NaClO + (NH₄)₂SO₄. A pilot-scale pipeline reactor was constructed and employed to study the overall TCR decay behaviors in response to the three kinds of disinfectant constituents. The contributions of the CI pipe to overall decay were quantified by using the newly developed refined total chloramine decay model (RTCDM). On the basis, the changes in pipe wall composition and corresponding TCR decay behavior were summarized for different disinfectant constituents. Finally, disinfection strategies were tentatively proposed for the chloraminated water supply system.

Chlorine and chloramine are active reagents and can participate in many biochemical reactions within the drinking water supply system. On the basis of the extensive research on disinfection chemistry, chloramine decay in water could be classified into bulk phase decay and pipe wall-induced decay. The bulk phase decay could be further ascribed to disinfectant autodecomposition and oxidizing reductive compounds, e.g., natural organic matter (NOM) and bioprotein. Every single reaction pertinent to chloramine autodecomposition has been fully characterized under different pH and temperature values (Vikesland et al. 2001; Huang 2008). The oxidation of NOM by chloramine and chlorine could be described as the biphasic second-order kinetic model with four specific reaction parameters (Duirk et al. 2005). On this basis, Wahman (2018) and Ricca et al. (2019) packaged the aforementioned two components into the prediction model to simulate TCR decay behavior in the batch reactor. The chloramine decay model and corresponding reaction rates for the prediction model could be obtained by reference to the previously published articles (Wahman 2018; Ricca et al. 2019). Besides, nitrite could also be oxidized by chloramine in the bulk phase (Margerum et al. 1994), which is more prominent in nitrified DWDS. Moreover, it has long been observed that pipe wall-induced chloramine decay accounts for much of the overall loss (Rossman et al. 1994; Huang & McBean 2008). As a result, we incorporated the nitrite oxidation component and pipe wall consumption component into the prediction model and developed the total chloramine decay model (TCDM) in our previous study (Ma et al. 2020). When used in a real distribution system, the TCDM returned satisfying prediction results on TCR decay behavior.

As for an aqueous solution, chloramine concentration is in chemical equilibrium with that of chlorine. The trace chlorine existing in water would manifest its significant effect on TCR decay when there is no pipe wall-induced consumption. It is found that chlorine experiences autodecomposition to produce or O₂, which is temperature, pH, and -N dependent (Adam et al. 1992; Adam & Gordon 1998). Therefore, the monochloramine hydrolysis reaction and chlorine autodecomposition model were incorporated into the original TCDM to better describe the TCR decay behavior in chloraminated water with trace chlorine. The RTCDM consisting of 23 differential equations for kinetic reactions and 4 algebraic equations for equilibrium reactions was edited in C programming language on MATLAB software (version R2018b) and solved using differential algebraic equations solver ode15s. The RTCDM can return the calibrated wall decay coefficient with its computing core using the nonlinear least square approximation method.

The wall decay coefficients were calibrated using RTCDM based on three sets of water quality data obtained from pipeline reactor experiments. As for the bulk phase decay experiments, the bulk decay coefficients were obtained by fitting the experimental data to the pseudo-first-order kinetic model, which was also edited in C programming language on MATLAB software (version R2018b). The curve fitting calculation would stop only until the size of the gradient is less than the value of the optimality tolerance.

To isolate the pipe wall consumption from overall TCR decay, bulk phase decay experiments were carried out by filling brown glass bottles with prepared feed water (NaClO, NaClO + NH₃·H₂O, NaClO + (NH₄)₂SO₄). Parallel experiments were conducted simultaneously for each feed water. Six to eight samples were collected from 0 to 48 h for each experiment. TCR, alkalinity, pH, water temperature, -N, -N, TOC, and were determined for each sample. The data used in model calculation are the average value of the two data from the parallel experiments.

On installation, the inner wall of CI pipes has to get dried with many tuberculations unevenly spaced in the pipe. Hence, we initiated the reactor on 15 April 2020 with treated water to reestablish the microbial community in the pipeline at the flowrate of 6.2 m³·h⁻¹, which is similar to that in DWDS. During the 83-day operation, the feed water was refreshed every 2 days. The formal TCR decay experiment was started on 7 July 2020 using fresh treated water. After a 3-day operation, the reactor was stopped, and the experimental water was drained out. Then the filtered water was fed into the system followed by the immediate addition of 2 mg·L⁻¹ NaClO (aq). As the pipe wall severely consumed chlorine disinfectant, TCR was increased to 2 mg·L⁻¹ by adding NaClO (aq) every 2 h. After a 24-h cleaning by free chlorine, the water was drained out and fresh filter water was fed into the system to conduct the formal TCR decay experiment with chlorinated water. After a 3-day operation, the water was drained out, and another 30-day operation was conducted by feeding fresh treated water every 2 days to promote microbes' adaptability to the chloraminated water once again. Subsequently, the last formal TCR decay experiment was carried out with fresh filtered water chloraminated by NH₃·H₂O for 3 days. The operational procedure of the pipeline reactor could be seen in Figure S1.

For every formal experiment, the reactor was operated under the same flowrate of 6.2 m³·h⁻¹. The sample was collected after the first 3 min of circulation and that time was set as the start point. The samples were collected at the operation time of 0, 1, 4, 8, 24, 48, and 72 h. TCR, pH, alkalinity, water temperature, -N, -N, -N, TOC, Fe residual, and were measured.

The treated water was obtained from Lingzhuang DWTP (Tianjin, China), adopting prechlorination (NaClO), coagulation, flocculation, settling, dual media filtration, and final chloramination by (NH₄)₂SO₄. The treated water was used directly as the NaClO + (NH₄)₂SO₄ water in the bulk phase decay experiments and pipeline reactor experiments. The filtered water was used as the NaClO water. NaClO + NH₃·H₂O was prepared by adding NH₃·H₂O into filtered water. The properties of feed water are presented in Table 1. TCR was measured by the N, N-diethyl-p-phenylenediamine method with a portable colorimeter (DR300, HACH, USA). pH was tested by a pH meter (S210-K, METTLER TOLEDO, Switzerland). Total alkalinity was determined by the acid–base indicator titration method according to the Water and Wastewater Monitoring Analysis Method. By using the UV–Visible spectrophotometer (DR6000, HACH, USA), -N was determined by the salicylate method at the wavelength of 655 nm with a resolution of 0.01 mg·L⁻¹, -N was determined by the diazotization method at the wavelength of 507 nm with a resolution of 0.001 mg·L⁻¹, -N was determined by the Cd reduction method at the wavelength of 507 nm with a resolution of 0.01 mg·L⁻¹, was determined by the Barium sulfate precipitation method at the wavelength of 450 nm with a resolution of 1 mg·L⁻¹, and Fe residual was determined by 1, 10-phenanthroline method at the wavelength of 510 nm with a resolution of 0.01 mg·L⁻¹. TOC was determined by the TC-IC method using a TOC analyzer (TOC-L, Shimadzu, Japan). The gross α and β radioactivities were measured by using evaporation, concentration, and enumeration methods.

Table 1

Properties of feed water used in the six sets of experiments

Parameters .	Bulk phase decay experiments .			Pipeline reactor experiments .
	NaClO + (NH4)2SO4 .	NaClO .	NaClO + NH3·H2O .	NaClO + (NH4)2SO4 .	NaClO .	NaClO + NH3·H2O .
Temperature (°C)	26.1	25.9	26.7	27.3	28.1	27.3
pH	7.730	7.959	7.928	7.620	7.826	7.719
Alkalinity (mg·L−1)	80	85	80	80	82	85
TCR (mg·L−1)	1.35	1.62	1.53	1.23	1.59	1.11
-N (mg·L−1)	0.44	0.00	0.40	0.47	0.00	0.33
-N (mg·L−1)	0.004	0.002	0.004	0.007	0.002	0.003
TOC (mg·L−1)	1.50	1.50	1.50	1.88	2.14	1.50
(mg·L−1)	35	33	32	35	33	32
Fe residual (mg·L−1)	Na	Na	Na	0.08	0.04	0.05
Gross α radioactivity (mBq·L−1)	30	31	32	21	21	22
Gross β radioactivity (mBq·L−1)	77	78	77	73	73	73

Parameters .	Bulk phase decay experiments .			Pipeline reactor experiments .
	NaClO + (NH4)2SO4 .	NaClO .	NaClO + NH3·H2O .	NaClO + (NH4)2SO4 .	NaClO .	NaClO + NH3·H2O .
Temperature (°C)	26.1	25.9	26.7	27.3	28.1	27.3
pH	7.730	7.959	7.928	7.620	7.826	7.719
Alkalinity (mg·L−1)	80	85	80	80	82	85
TCR (mg·L−1)	1.35	1.62	1.53	1.23	1.59	1.11
-N (mg·L−1)	0.44	0.00	0.40	0.47	0.00	0.33
-N (mg·L−1)	0.004	0.002	0.004	0.007	0.002	0.003
TOC (mg·L−1)	1.50	1.50	1.50	1.88	2.14	1.50
(mg·L−1)	35	33	32	35	33	32
Fe residual (mg·L−1)	Na	Na	Na	0.08	0.04	0.05
Gross α radioactivity (mBq·L−1)	30	31	32	21	21	22
Gross β radioactivity (mBq·L−1)	77	78	77	73	73	73

^aFe residual was not measured for the corresponding feed water.

View Large

For the bulk phase decay experiments, only the time series data of TCR were input into the pseudo-first-order kinetic model to conduct parameter (k_bulk) calibration. As for the pipeline reactor experiments, seven parameters listed in Table 1 (except Fe residual, ⁠, Gross α radioactivity, and Gross β radioactivity) were set as the initial conditions in RTCDM. The time series data of TCR were also input into the RTCDM to achieve parameter (k_wall) calibration. Upon model calculations, the effects of pipe wall and disinfectant constituent on TCR decay could be quantitatively reflected by comparing k_bulk and k_wall.

The TCR decay coefficients obtained from model calibrations were tabulated in Table 2. The bulk decay coefficient k_bulk for the two chloramination experiments were all 0.004 h⁻¹, while the coefficient for the chlorination experiment was 0.011 h⁻¹, a little bit larger than that of the former two. By calibrating the experimental data with RTCDM, the pipe wall-induced TCR decay was quantitatively isolated from the overall decay within the pipeline reactor. The comprehensive wall decay coefficient k_wall for the three pipeline reactor experiments was 0.465–0.864 h⁻¹. To evaluate the contribution to the overall decay, the ratio k_wall/k_bulk was calculated and shown in the last column in Table 2. The ratio ranged from 42 to 216; therefore, more than 97% of the overall decay could be explained by pipe wall consumption. In 2009, a pipe section reactor study, also using tuberculated cast iron pipe, calculated the k_wall of 0.597–0.709 h⁻¹ (Westbrook & Digiano 2009), which is within the boundary of our calibration results. However, our recent field study on the Tianjin distribution network found that the k_wall was evidently smaller than the results presented here, and the ratio was only about 5.0–9.0, which depended on the season and distribution routine (Ma et al. 2020). These outcomes manifest that identical numerical results could hardly be obtained as the bench- and pilot-scale studies are easier to keep the physical and chemical conditions constant, which is not the case for the field study (variations in pipe material, water velocity). However, no matter for which scale, all these studies proved that the pipe wall could dominate the overall chloramine decay in the pipes, especially for the highly corrosive cast iron pipes.

The contribution of microbe consumption and electrochemical corrosion to the TCR decay was not quantified in this study. But, the two pipe wall-relating factors have shown their significant effects on the chlorine/chloramine decay process, which is much stronger relative to the bulk decay. Given the highly corrosive pipe attached by well-developed biofilm, how the different compositions of disinfectants react to the pipe wall needs to be further addressed.

In a real distribution system, different compositions of disinfectants are used. Here, NaClO, NaClO + (NH₄)₂SO₄, and NaClO + NH₃·H₂O were studied to compare their variation behaviors in bulk water and cast iron pipeline systems. As shown in Table 2, the bulk decay coefficients are the same for the two chloramination experiments. So the origins of -N ((NH₄)₂SO₄ or NH₃·H₂O) would not obviously influence disinfectant decay dynamics in the bulk phase. Comparatively, the decay coefficient for NaClO disinfection is 2.75 times that of the former two. The accelerated decay behavior is the result of the faster autodecomposition of chlorine (Kulkarni et al. 2018) and the faster oxidation with NOM (Duirk et al. 2005) in the aqueous solution. Interestingly, the wall decay coefficient for NaClO disinfection is 0.465 h⁻¹, which is the smallest among the three experiments (Table 2). However, the decay coefficient for NaClO + NH₃·H₂O disinfection is 1.55 times that for free chlorine disinfection. The decay process further accelerated NaClO + (NH₄)₂SO₄ disinfection with a decay coefficient 0.86 times higher than that for chlorine disinfection. The reverse order in decay coefficient manifests the different reactions of the pipe wall to the disinfectant constituent, which could further lead to the different disinfectant decay behavior.

Even chloramine could better penetrate into the biofilm (Lee et al. 2011), -N produced upon inactivation or oxidation could serve as a nitrogen source to nitrifier and other bacteria (Hua et al. 2011), which may significantly boost bacteria reproduction within the pipeline. Comparatively, chlorine has the better microorganism inactivation ability and could effectively kill bacteria and prevent CDPs from producing and releasing by bacteria. As a result, the bacteria-mediated TCR decay could be directly slowed down upon free chlorine disinfection. What is more, it is reported that nitrification could remarkably decrease pH within the biofilm (Pressman et al. 2012). Under the low pH condition, Fe oxidizing bacteria could utilize Fe²⁺ to produce Fe³⁺, which may speed up the iron corrosion process (Teng et al. 2008). On the other hand, it is found that free chlorine disinfection could stimulate iron release from the pipe wall obviously relative to chloramine disinfection (Li et al. 2014; Hu et al. 2018). Therefore, corrosion-relating TCR decay observed here was mainly attributed to the microorganism-mediated pathway. The different reactions of heavily polluted pipe wall to NaClO and NaClO + NH₃·H₂O cause the difference in the TCR decay behavior.

The introduction of -N could stimulate the development of nitrifiers and other microbes within the attached biofilm. The additional would further exacerbate the TCR decay problem by encouraging SRB and other microbes' reproduction and metabolism. Luckily, free chlorine disinfection could be used to cope with the problem with higher performance relative to combined chlorine disinfection.

Based on the aforementioned discussion, chlorine was found to be suitable for slowing down TCR decay and limiting microorganism development within the highly corrosive pipeline system. The introduction of -N or has an adverse effect on the maintenance of the TCR level. Therefore, it is supposed to minimize and addition to drinking water delivered by the highly polluted cast iron pipe.

With the advance of technology, cement-lined ductile iron pipe is found to be inert in water chemistry with high strength. The cement-lined pipe is reported to have fewer effects on TCR decay for chloraminated water (Westbrook & Digiano 2009). Therefore, the long-used cast iron pipes are being replaced by the cement-lined pipes in recent years. The changes in pipe materials can reduce the disinfectant consumption in water mains. But the small diameter CI pipes left in community or buildings would still lead to the sharp drop in TCR for tap water, especially during the hot season (Ma et al. 2020). Besides disinfectant lost in DWDS, production safety may also affect decision-making on disinfection craft. For example, free chlorine and are dosed in the form of NaClO and (NH₄)₂SO₄ instead of Cl₂ and NH₃·H₂O, respectively, in some large-scale water supply systems in China (e.g., Beijing and Tianjin). In fact, the disinfection strategy is the result of the trade-off among water quality safety, practical feasibility, and low cost.

It should be noted that the evident biotic- and abiotic-mediated TCR decay has been broadly observed under the high water temperature, but it is not the case for the low water temperature. The retarded biochemical reaction under low temperature would be more compatible with chloramine disinfection. So a flexible disinfection strategy may be more suitable for the specific water supply system. The consistent free chlorine disinfection may be suitable for the small-scale DWDS or the DWDS having a large proportion of long-used CI pipe. For large-scale water supply systems, combined chlorine disinfection can be used to maintain TCR level throughout the DWDS, on the condition that only limited CI pipes are still in use. In that case, NH₃·H₂O, instead of (NH₄)₂SO₄, is recommended to accomplish the chloramination to minimize the risk of stimulating microorganism regrowth. But if the water temperature is in transition to a high level, measures should be taken in advance to prevent microorganism boosting in DWDS. Given microbial problems typically are worse when pipeline water is 15 °C or higher (Neden et al. 1992), DWTP is supposed to switch to free chlorine disinfection to perform pipeline cleaning before the critical temperature comes. After which, DWTP can reuse the combined chlorine as the secondary disinfectant. This strategy has been adopted by some water supply systems, such as Pinellas county utilities in America (Hua et al. 2011) and Foshan city utilities in China (Deng et al. 2015). Overall, to constrain microorganism activity in chloraminated DWDS, it is recommended to reduce free ammonia levels in the entry-point by using high Cl₂:NH₃ ratios, carry out free chlorine maintenance timely by DWTP, replace the aged CI pipes, and flush the distribution pipes of low velocity. The usefulness of free chlorine maintenance has been substantiated for the highly corrosive pipes in this pilot study. Interestingly, NH₄Cl may serve as the optional composite to accomplish the treated water chloramination with the advantage of no extra inorganic salt introduction. Hence, its effectiveness on distribution pipes, especially for the aged CI pipes, could be evaluated in the future to see its advantages over (NH₄)₂SO₄.

The pipeline reactor was employed to study the effects of highly corrosive cast iron pipes on the disinfection chemistry at the pilot scale. The contributions of CI pipe under different compositions of disinfectants were quantified by using the refined TCDM. The key conclusions are as follows:

• 1.

  The newly developed RTCDM could well simulate TCR decay behavior based on the experimental data and give the calibrated k_wall to quantify the pipe wall contributions.

• 2.

  NaClO + (NH₄)₂SO₄ may promote microorganism regrowth and metabolization in CI pipe with unevenly spaced tuberculation, which further accelerates the TCR decay process. Comparatively, NaClO + NH₃·H₂O has the features of a smaller k_wall as no extra introduction. Free chlorine could effectively inactivate or kill microorganism attached to the pipe wall, and as a result, the overall TCR decay process could be slowed down.

• 3.

  The forces driving TCR decay in highly corrosive CI pipe include iron-mediated electrochemical corrosion and microbe-mediated (e.g., nitrifier and SRB) biotic disinfectant consumption.

• 4.

  It is recommended to minimize extra inorganic salt introduction into treated water to constrain microbial development in DWDS. As for the chloramine disinfection system, DWTP could conduct free chlorine maintenance to clean the whole DWDS. Besides reducing the free ammonia level, replacing and flushing the distribution pipe may be beneficial for maintaining the TCR level.

The authors acknowledge the funding offered by Tianjin Water Group Co. Ltd (Project Number: 2019KY-02). The authors would like to acknowledge PE. Fushou Wan of Tianjin Hongyuan Water Treatment Engineering Co., Ltd. for his help in providing us the insight into hydrodynamics. The authors also acknowledge Vice Manager Feng Xiao of Lingzhuang Water Treatment Plant in Tianjin for his help in providing the filtered and treated water.

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

The authors declare there is no conflict.