Residual Chlorine Simulation in Water Distribution Network T.Kondo 1, T.Fuchigami 2 1 Water Distribution Department of Osaka City Waterworks Bureau,14-16-1 Nankou-Kita, Suminoe-ku, Osaka-city, Osaka, JAPAN,(E-mail: t-kondou@suido.city.osaka.jp) 2 Water Examination Laboratory of Osaka City Waterworks Bureau,1-3-14 Kunijima, Higashiyodogawa-ku, Osaka-city, Osaka, JAPAN Annotation Osaka City is working to maintain free residual chlorine concentrations as low as possible but within the range of the required legal standard, at the end of the water distribution network in order to improve the efficiency of disinfectant chlorine injection and reduce chlorine odor in tap water. We have developed a simulation method as a means for achieving these goals, to directly estimate free residual chlorine concentrations at arbitrary points in the water distribution network in Osaka City using the first- and zero-order rate equations for the reduction in free residual chlorine concentration. We have also conducted two case studies to compare the measured values with simulated ones. Consequently, a good reproducibility was confirmed. We intend to utilize our simulation method not only to control free residual chlorine concentration properly in distribution network but also to detect the points where free residual chlorine concentration becomes too low level because water keeps staying, which leads to identify the effective pipe renovation work from the aspect of the reduction of free residual chlorine decay. Keywords Free residual chlorine, water distribution network, simulation of the reduction rate of free residual chlorine concentration, EPANET2, rate coeffi cient for the reduction rat e of free residual chlorine concentration 1. INTRODUCTION The control of free residual chlorine concentrations in the water distribution network is an essential task to ensure the safety of tap water against infectious microorganisms. The Japanese Waterworks Law stipulates that free residual chlorine in tap water is kept at a concentration of 0.1 mg/l or greater, however, the results of a customer survey of inhabitants of Osaka City showed their desire to reduce free residual chlorine concentrations as much as possible for reduction of tastes and odors cause by chlorine while maintaining the safety of the tap water. Free residual chlorine concentrations vary between areas surrounding a water purification plant and water distribution plants, where chlorine is put into the water and areas at the end of the water distribution network due to different flow times. At present, to maintain free residual chlorine concentrations at the end of the water distribution network at 0.1 mg/l or greater, the Osaka City Waterworks Bureau (OCWB) aims to restrict the upper concentration to 0.4 mg/l. However, to supply water with high quality in terms of odor and taste and to reduce the cost of chlorine injection, it is effective to reduce the differences in free residual chlorine concentrations in Osaka City and maintain them as low as possible, therefore, the OCWB is installing additional chlorine injection facility in secondary water distribution plants. To control chlorine injection, the OCWB sets the volume of chlorine injection at the output of the water purification plant and in some of water distribution plants installed with facilities for additional chlorine injection, in order to maintain free residual chlorine concentrations at the end of the water distribution network at 0.1 mg/l or greater. The OCWB is monitoring the quality of water
in distribution pipes at 38 sites in Osaka City using water quality sensors in order to prevent free residual chlorine concentrations from falling below the lower limit. Furthermore, if a simulation method to accurately estimate a free residual chlorine concentration at an arbitrary point in the water distribution network in addition to the above measures is available, it is allowed to control free residual chlorine concentrations in broader areas in a precise manner, i.e., optimal chlorine injection including additional injection and appropriate measures for local reduction in free residual chlorine concentrations of some areas due to the status of the water distribution network (water retention and chlorine consumption on the inner surface of aging pipes). The OCWB started to develop a simulation method in 2007 and conducted a case study including a comparison between simulated and measured concentrations in two water distribution areas in Osaka City. Consequently, a good reproducibility of simulation results was confirmed. The process is described below. 2. METHODS 2.1. Development of a distribution network model The analysis software used was EPANET2 [1], a program that models water flow including mixing and separating of water flows in water distribution pipeline network and provides time series data analysis, integrated with a geographic information system. A distribution network model was developed using the geolocation and pipe line profile and demand data obtained from the pipeline information system that was organized using GIS by Osaka City, as well as data on ground height in digital national land information provided by the Geographical Survey Institute. 2.2. Rate equation for the reduction in free residual chlorine concentrations The simulation method for concentrations of free residual chlorine in a water distribution network accurately estimates the flow time at an arbitrary point in the network using time series analysis, considering changes in water demand and mix and separate flows at nodes, and directly estimates free residual chlorine concentrations using the first- and zero-order rate equations for the reduction in free residual chlorine concentrations. Considering a previous study [2] and report [3] of the OCWB, the following equation is used as a rate equation for the reduction in free residual chlorine concentrations in water distribution pipes. dc dt 4 ( kb C kw) (1) d where, C: free residual chlorine concentration (mg/l) k b: bulk decay rate coefficient for the reduction in free residual chlorine concentration (day -1 ) k w : rate coefficient for the reduction in free residual chlorine concentration derived from the inner surface condition of a water distribution pipe (mg/m 2 /day) d: diameter of a pipe (m) 2.3. Conditions in time series analysis for water distribution networks Hydraulic calculation was performed using the Hazen-Williams equation, and the roughness constant for pipe walls in the equation (C) was specified as 90 in all cases. As for the effluent volume at the output of each water distribution plant and free residual chlorine concentrations at that point, which are the start points of the analysis, the data for 24 hours from the data sequentially monitored and measured was given as the default condition. The simulation method for free residual chlorine concentration was assessed by comparison between measured free residual chlorine concentrations in water samples collected from a fire
2.4.2. Validation in the water distribution area of Nagai Water Distribution Plant (Case study 2) In other water distribution areas in Osaka City except for those of Sakishima Water Distribution Plant, various pipes with differences in the type of inner lining and the year of installation had been laid. Therefore, to extend the application of this simulation method, the second case study was conducted in a water distribution area with plural type of pipes in order to assess the reproducibility of this simulation method considering differences in pipe types. Based on the results of preliminary studies, the rate coefficient for the reduction in free residual chlorine concentration by distribution route was estimated considering the year of installation and the type of inner lining of distribution pipes. Subsequently, after substituting the estimated rate coefficient into the rate equation for the reduction in free residual chlorine concentration, the effect of the differences in pipe type was evaluated. Overview of the water distribution area of Nagai Water Distribution Plant and distribution pipes laid down in the area Nagai Water Distribution Plant is located in the southeastern region and distributes water by pump pressure to surrounding areas which have approximately 125,000 households. This area is isolated from surrounding distribution areas by shutting valves installed at the boundaries between distribution areas. The maximum distribution volume in this area is approximately 110,000 m 3 /day. As for the water-supply system, water is distributed from Niwakubo Water Purification Plant to Tatsumi Water Distribution Plant, the primary plant, using a transmission pump. Subsequently, water pooled in a reservoir in Tatsumi Plant is distributed to Nagai Water Distribution Plant, the River Yodo Niwakubo Purification Plant Chlorine injection Tatsumi Distribution Plant Nagai Distribution Plant secondary plant, using a water-distribution pump, while some of water is distributed to households (see Figure 3). Chlorine in water distributed from Nagai Water Distribution Plant is injected in two stages, i.e., in the final process of water purification at Niwakubo Water Purification Plant, and subsequently at the entry of Nagai Water Distribution Plant as an additional injection. However, at present, chlorine injection in the output of Niwakubo Water Purification Plant is substantial in order to maintain free residual chlorine concentrations at 0.1 mg/l or greater, because, in Niwakubo disutribution area, secondary distribution plants except Nagai Water Distribution Plant are not installed with facilities for additional chlorine injection and the free residual chlorine concentration of water supplied from Nagai Water Distribution Plant is sequentially monitored using two water-quality telemeters installed in this distribution area. The total length of pipes in the water distribution area of Nagai Water Distribution Plant is 330 km, and pipes with a small-diameter of 300 mm or less accounts for approximately 89%. The pipes used comprised non-lined cast-iron pipes, cast-iron pipes renovated with liquid epoxy resin-lining, and mortar-lined ductile iron pipes. Mortar-lined ductile iron pipes that were laid down in 1997 or before accounts for approximately 84% of the total pipe length(see Figure 4). P P Tatsumi supply area Additional chlorine injection Nagai suppl y area Figure 3 Water distribution system for Nagai Water Distribution area P
The inner surface type of distribution pipes Mortar-lined (after 1998) Mortar-lined (before 1997) Epoxy resin-coated (renovated pipes) Non-lined Nagai Water Distribution Plant 3. 8% 6.2% 0. 5% 5.1% 12.2% Total length of pipes 330km 88.7% Pipes by diameter 300<D 300<D 500 500< D Ratio of distribution pipes by diameter Total length of pipes 330km 83. 5% Pipes by inner surface type Liquid epoxy res in-coated Mortar-lined (1998- ) Mortar-lined (-1997) Non-lined Ratio of distribution pipes by inner-surface type Figure 4 Overview of pipe network in Nagai Water Distribution area Simulation of pipe network in Nagai Water Distribution area The k b value in the above Equation (1) was fixed separately [5], similarly to Case Study 1, using the following equation (Equation (2)), an approximate equation obtained from the laboratory results with water treated in a water purification plant in Osaka City, by substituting water temperature on the measurement day. k b 0.0408exp(0.0742T ) (2) where, T: water temperature ( C) In comparison between measured and estimated concentrations, the correlation coefficient between them was 0.81 (n=21) and the profile of reduction in free residual chlorine concentrations was approximately reproduced. Consequently, the application of this simulation method is considered to be acceptable. In addition, the water demand of each household was set at a nodal point nearby the pipe network model using the annual data of water consumption by water supply meter for approximately 125,000 households. Since common houses and apartments have been built in most of the distribution area of Nagai Water Distribution Plant, an hourly change in water demand patterns over the whole area of Nagai Water Distribution Plant corresponds to an hourly change in patterns of total water distribution amounts at Nagai Plant (see Figure 5).
Nagai Wat er Distribut ion Plant 7000 0.70 Overview of the distribution network model Number of pipes: 2546 Number of joints: 3847 Hydraulic calculation: Hazen-Williams Equation C value: 90 Water distribution volume (m 3 /h) 6000 5000 4000 3000 2000 1000 1:00 3:00 5:00 Average water temperature (effusion site ): 29.0 C Da ily total volume of wa ter distribution 106, 000 m 3 (data a s of Se ptember 13, 2007) 7:00 Tota l volume of wate r distribution from Nagai Water Distribution Plant (m 3 /h) Initial residual chlorine concentration (mg/l) (effusion site of Nagai Water Distribution Plant) 9:00 11: 00 13: 00 Time Volume of water distribution and initial residual chlorine concentration in Nagai Water Distribution Plant Figure 5 Overview of the distribution network model in Nagai Water Distribution area Based on the results of time series analysis for Nagai water distribution network in advance, candidate sites with a low concentration of free residual chlorine were selected, and subsequently, free residual chlorine concentrations were measured at 21 fire hydrants that were located at these sites or the distribution routes which lead to the these sites(see Figure 6). Simulation results using k w values based on analysis by pipe type was compared with that based on analysis with a specified constant as k w value, in order to validate the effect of simulation using k w values by pipe type. 15: 00 17: 00 19: 00 21: 00 23:00 0.65 0.60 0.55 0.50 0.45 0.40 Residual chlorine concentration (mg/l) The inner surf ace type of distribution pipes D Yamasaka N=3 Mortar -lined (after 1998) Mortar -lined (before 1997) Epoxy resin-coated ( renovated pipes) Non-lined Nagai Water Distribution Plant H Minamisumiyoshi E Nagaihigashi G Terugaokayata C Nagayoshirokutan N=2 N=2 N=3 N=1 N=3 N=2 A Uriwarihigashi N=2 N=3 B Nagayoshi kawanabe F Yamanouchi Figure 6 Site map of fire hydrants for free residual chlorine concentration measurement
Measurement survey of rate coefficients for the reduction in free residual chlorine concentration derived from inner surface status by pipe type(k w ). The k w values were measured by pipe type in a measurement survey. Of distribution pipes (diameter: 100 to 300 mm) in the distribution area, pipes in which water can be pooled intentionally by shutting valves without affecting customers (e.g. pipes which are a parts of network loop and have no service connection) were selected and after measuring the initial concentration of free residual chlorine, the time to reach the concentration of 0.10 mg/l or less, a free residual chlorine concentration at that point and water temperature in a water sample were measured. First, the k b value was estimated using the above Equation (2), and subsequently substituted into Equation (1) with other measured data. The k w values are shown in Table 1. Table 1 Measurement results of k w by pipe type Pipe type kw value Rati o of pi pes in water distribution area (mg/m 2 /day) (%) High-quality cast iron pipe / Non-lined pipe 705.6(n=1) 0.5% High-quality cast iron pipe / Liquid epoxy-lined pipe 9.6(n=3) 3.8% Ductile iron pipe (before 1997) / Mortar-lined pipe 11.5(n=6) 83. 5% Ductile iron pipe (after 1998) / Mortar-lined pipe 4.3(n=4) 12. 2% 3. RESULTS AND DISCUSSION Figure 7 shows the comparison between measured and estimated concentrations when different k w values by pipe type were substituted. On the other hand, Figure 8 demonstrates the comparison when k w value was constant (used a k w value of mortar-lined (old) pipe, which was most of the total length of distribution pipes). The correlation coefficient of the former was 0.818 while that of the latter was 0.790. Since both of them were high, the profile of reduction in free residual chlorine concentrations was almost reproduced. When k w values classified by pipe were used, the reproducibility of simulation outcomes increased compared with that of simulations using a constant k w value. In particular, reproducibility was improved in the end area of the southeastern region, in which most pipes laid were relatively new. In this simulation, accuracy was improved by approximately 10% in the standard deviation of errors between measured and estimated concentrations. Therefore, in this water distribution area, where the majority of pipes was mortar-lined ductile iron pipe and laid down before 1997, the accuracy of simulation outcomes was considered to be sufficiently acceptable in practice when the water flow time in pipe networks was accurately reproduced, even if the k w value of the major pipe type was used for all of the pipes. 0.00 0.00 0.10 0.20 0.30 0.40 0.50 0.60 Theoretical value Figu re 8 Comparison between measured and estimatedconcentrations when kw was constant However, the k w value of non-lined pipe was markedly faster than that of other type pipes, therefore, in an area distributed through many non-lined pipes, simulation with k w values classified Actual measurement Actual measurement 0.60 0.50 0.40 0.30 0.20 0.10 R=0.82 0.00 0.00 0.10 0.20 0.30 0.40 0.50 0.60 Theoretical value 0.60 0.50 0.40 0.30 0.20 0.10 Residual chlorine concentr atio n (mg/l) Correlation coefficient Figure 7 Comparison between measured and estimated concentrationswhen kw was classified by pipe type Resid ual chlorine concentration (mg/l) Correlation coefficient R=0.79
by pipe type should be conducted. In the k w measurement survey, water sampling was carried out at 12 points due to the limitation of available pipe routes. Since the k w value of non-lined pipes was markedly faster than that of other type pipes, when long-term water pooling in non-lined pipes is predicted by the simulation outcomes, the relevant pipes should be replaced earlier and regular monitoring and control systems for free residual chlorine concentration in seasons with high water temperature should be established. 4. CONCLUIONS The results of the two case studies validated a good reproducibility in the simulation outcomes of free residual chlorine concentration, and therefore, this simulation method developed by the OCWB is expected to be applied under the following conditions: to estimate target values of chlorine injection optimal for each water purification plant and distribution plant when separate injection of chlorine is installed. to predict an area where free residual chlorine concentration can be reduced by water pooling in distribution pipes and plan a reasonable renovation of pipe networks for water quality control, considering the reduction properties of free residual chlorine concentration from the standpoint for the reduction of pipe diameter and the priority of renovation, and to examine whether water pooling occurs in distribution network due to changes in water flow associated with long-term water suspension of a part of the network due to renovation construction work. Future tasks are to improve the accuracy of simulation outcomes and expand versatility in order to apply the measures to all of Osaka City. To improve the accuracy of simulation outcomes, it is important to evaluate the properties of pipes such as C values in a distribution network model, and improve the reproducibility of water flow conditions including the use of the measured values of water pressure and flow amounts in distribution pipes, discharge pressure and flow amounts in water purification and distribution plants. In addition, to expand the application of the simulation method to a practical level, sensitivity analysis should be conducted on the effects of changes in variables including water temperature, k b, k w, and C values on simulation outcomes. It is also important to judge to what extent accuracy should be improved in the future. To expand application of the simulation method to all of Osaka City, a distribution network model including border areas where water flow turns between adjacent distribution blocks should be developed and organized as a model with water flow patterns between these blocks. Considering the above issues, we intend to improve the accuracy of this simulation method and expand its versatility. REFERENCES [1]Rossman, L. A. (2000). EPANET2 Users Manual.EPA-600/R-00/057. USEPA National Risk Management Research Laboratory, Cincinnati, USA [2]FUCHIGAMI, Tomohiro and TERASHIMA, Katsuhiko. (June 2005). Development of control method for residual chlorine concentrations considering chlorine consumption on the inner surface of distribution pipes. Journal of the Japan Water Works Association, vol.74, No.6 (No.849), pp. 15-26 [3]John, J.V.et al. (1997). Kinetics of chlorine decay. JOURNAL AWWA, Vol.89, pp.54-65. [4] NAGATANI,Toru. et al. (2006). Residual chlorine concentration simulation in water distribution network. The 7th International Symposium on Water Supply Technology [5]FUCHIGAMI, Tomohiro and TERASHIMA, Katsuhiko. (2003). Behavior of residual chlorine in water aft er advanced puri fication treatment in distribution pipe process in Osaka City and its control. Journal of the Japan Water Works Association, vol.72, No.6 (No.825), pp.12-24