Water Quality Modeling under Gradually Varied Flows in Distribution Systems

Abstract

Most of the real water distribution systems have a condition of gradually varied flow when nodal demand or tank level varies over time. Unlikely prevailed water quality models, the developed model predicts the flows and chlorine residuals of pipe networks through transient analysis based on the rigid-column assumption. Discrete volume method was applied to solve advective transport equations in pipes and constituent reaction is included at the same time. The developed model was applied to two pipe networks and the results were compared with those of EPANET and EPANET2. The results show that the inertial forces have an effect on the change of flow in pipes and EPANET and EPANET2 may have some faults in calculating the chlorine concentration of tank in the codes. Introduction The objective of a drinking water distribution system is to deliver sufficient quantities of water where and when it is needed at an acceptable level of quality. In the past, the water distribution system was designed and operated to meet hydraulic performance, but in recent years, water quality has been a major concern of water authorities since an increased number of incidents of violation were reported (Clark and Grayman, 1998). As water travels through the distribution system, the quality of treated water deteriorates due to the decay of disinfectant and the regrowth of microorganisms, in particular, in case of long residence time. Various numerical models have been presented for predicting the spatial and temporal distribution of chlorine residuals in distribution systems (Liou and Kroon, 1987; Rossman et al., 1993; Boulos et al., 1995; Islam and Chaudhry, 1996). Most of these models assume the flow to be steady, although some employ the extended-period simulations (EPS) in which a sequence of steady-state simulations are carried out by updating the tank levels and introducing nodal demand changes. But most of the real water distribution systems have a condition of gradually varied flow when nodal demand or tank level varies over time. Thus the existing models may not produce reliable results in the real applications. Water-hammer analysis or distributed system approach may calculate the flow changes in a network accurately but do not justify the costly use in gradually varied flow conditions (Karney, 1984; Holloway, 1985). On the other hand, the EPS technique does not adequately represent the flow and pressure variation within the system since the inertial effects are neglected. The lumpedsystem approach gives satisfactory results with far less computational efforts than that required by the water-hammer analysis (Holloway, 1985). In the lumped-system analysis, the inertial forces are included while the compressibility effects of both fluid and pipe walls are neglected. With the obtained flows in the network, the transport equation is numerically solved by using discrete volume method (DVM) suggested by Rossman et al. (1993). Hydraulic Modeling Governing Equations. The mathematical representation of the flow of a compressible fluid in pipes generally requires two of the three basic equations: (i) conservation of mass, (ii) conservation of energy, and (iii) conservation of momentum (Karney, 1984). Conservation of mass states that the rate of storage in a pipe system is equal to the difference between the inflow and outflow to the system. For a junction node,