Drinking water distribution networks (WDNs) are vulnerable to contamination because of their ubiquity, multiple points of access, and aging infrastructure. Looped characteristics and time varying flow patterns introduce additional challenges to identify the best response management strategy once the network is contaminated. Though intentional contamination of WDNs may be a relatively rare event, it is still important to prepare emergency responses to help mitigate the consequences and provide public assurance. The main drawback of the existing literature in consequence management is failure to consider pressure deficit conditions in the network which calls for an adaptive strategy, as system topology may change dynamically by implementing any new solution.;This research has two parts. The first part presents an optimization model to position water quality monitoring stations in a pressurized water distribution system (WDS). The optimization model is formulated as an integer program, and the solution of the mathematical problem is efficiently approximated using a multiobjective multi-colony ant algorithm. A built-in routine is developed for calculating the water fraction matrix and integrated into the general modeling structure to facilitate data entry and storage to minimize problems of water fraction matrix determination for varying scenarios and coverage criteria for any scenario. The proposed method is robust in analyzing the effects of different scenarios and/or number of potential monitoring stations by eliminating the need of employing an off-line routine for coverage matrix identification. Robustness, ease of generalization, multiobjective nature, and computational efficiency are the main characteristics and novelty of the proposed approach.;The second part presents an integrated simulation-optimization scheme for operational response management in an intentionally contaminated network. During consequence management, system topology may change from one decision point to another by varying valves and hydrants modes of operation. Modified topology may cause large negative pressures at different nodes if pre-specified demands are forced to be satisfied, which is exactly what happens in demand driven analysis (DDA). Discussing the drawbacks of demand driven network solvers in consequence management strategies, it integrates a pressure driven network solver (PDNS) with the multiobjective honey bees' mating optimization (HBMO) algorithm to develop single period and multi-period strategies. This study develops single and dynamic (multiple period) strategy. The dynamic strategy adjusts the decision at each stage per prevailing conditions. It accounts for time variation of demand at demand nodes and changes in flow direction and flux in new network topology from previous actions.;Solutions to the proposed multiobjective optimization problem generate a set of non-dominated optimal operational strategies which minimizes consequences of intentional physical attacks on water infrastructure systems. Each trial solution developed by the optimizer defines a new network topology due to modified modes of operation of nominated valves and hydrants. The PDNS finalizes the nodal pressure and modifies nodal withdraw for the identified trial solution. The PDNS is a modified version of the EPANET which handles changes in network topology while simulating networks with a pressure-deficit condition. Performance of the single-period and multiple-period modeling approach is illustrated using previously tested examples. It is shown that proposed multiple period strategy may reduce the negative impacts to public health, however, the computational time may introduce a new challenge in large networks.
展开▼
