## ABSTRACT

The paper introduces the Variable Speed Pump coefficient direct computation that has been implemented in the latest water distribution modelling software. The Water distribution simulation packages are often based upon EPANET hydraulic solver that has been developed by employing the Global Gradient Algorithm (GGA). Originally, the computation of the Variable Speed Pumps (VSP) coefficient was an iterative procedure, which could lead to convergence problem. In order to overcome this problem, a direct computation algorithm of the VSP coefficient was incorporated into the GGA, giving rise to a complex non-symmetric problem, which was solved by partitioning the original system matrix to lead to the solution of a large symmetrical problem, which size remains more or less the number of demand driven nodes, plus a small non-symmetrical problem, which size equals the number ofVSPs. The results of the approach were tested on the benchmark examples. After exhaustive debugging and testing, the algorithm was finally introduced in WaterCAD and WaterGEMS for the purposes of water distribution system analysis.

Water supply systems frequently present high-energy consumption, which correspond to the major expenses of these systems. Energy costs are a function of real consumption and the daily tariff. This paper presents a model of multi-criteria optimization for energy efficiency in water supply systems. The system is equipped with a pump station and presents excess of available energy in one of its gravity branches. An optimization method to define the best pump operation planning along the simulation time period, while satisfying the system constraints and population demands, is implemented in order to minimize the operational costs. The model, developed in MATLAB, provides the best solution to take in each time step. The rules are subsequently introduced in a hydraulic simulator (e.g. EPANET), to verify the system behaviour along the simulation period. The results are compared with the normal operation. The introduction of a water turbine generates not neglected economical benefits.

The current paper focuses on the analysis and optimization of operating strategies of water supply systems taking into consideration different points of view related to the technical-hydraulic and Water quality performance of the system. An integrated software tool has been developed composed of three main modules: a hydraulic simulator, a performance assessment tool and an optimization module based on Genetic Algorithms (GA). The developed tool has been applied to a real life case study: a water conveyance system — “Póvoa Vila do Conde” system - integrated in the Multimunicipal Water Supply System of “Baixo Cávado” and Ave. Results have shown that the implementation of optimal pumping schemes allows the reduction of operation and maintenance direct costs associated with pumping energy between 6 to 8% as well as the increase of hydraulic reliability of the system, maximizing available water volume in storage tanks,

The process of establishment and making solution of pump optimal control model was presented based on simplified model of large-scale water distribution network. When the simplified model was used instead of the microscopic model, the calculation time of optimal control model was greatly saved, but the precision was basically the same. The results of optimal control model for the water distribution system in Tianjin city showed that the simplified model for optimal control had remarkable advantage,

This paper summarizes the hydraulic network modeling study to assess the existing and future storage and pumping capabilities of the Chilliwack Mountain area (Chilliwack, BC, Canada). Development in the Chilliwack Mountain area will be significant in the next 15 years. This study determined that the current storage and pumping capacities of Chilliwack Mountain will not be able to support these new developments. Infrastructure upgrade strategies are developed and modeled to determine the most cost-effective way to eliminate the deficiencies in the Chilliwack Mountain water sy stem over the next 15 years. Improvement projects including storage tank, pump, valve and pipe upgrades are recommended. This study demonstrates that a complex hydraulic network model can be used as a practical tool for the optimal planning of network upgrades to meet fire flow requirements. Enhancement of water distribution infrastructure planning, operation, and management are principal benefits to the city of this study.

Case studies show that it is possible to make significant energy savings by calculating optimal pump schedules for two large-scale, water distribution systems while satisfying pressure limits. The method takes into account the pump head-flow characteristics, electrical tariffs and reservoir levels. Optimal pump schedules: can lead to new rules of operation with reference to the reservoir levels and energy tariffs.

The problem of least cost design of Water Distribution Systems (WDS):is defined as the constrained optimisation problem with the objective being the minimisation of total design/rehabilitation costs where the constraints are the minimum required pressure heads at all WDS nodes, mass and energy balance equations and decision variable search bounds. Decision variables are the design/rehabilitation options available for each WDS. The problem is solved here using the hierarchical Bayesian Optimisation Algorithm, hBOA (Pelikan & Goldberg, 2000), a probabilistic model building genetic algorithm that uses a Bayesian network to model the set of promising solutions in each generation and in turn samples this network to generate the offspring to be incorporated into the next generation. hBOA has recently been tested on two benchmark least cost design WDS problems; the New York Tunnels problem and the Anytown network (Olsson et al., 2007). The results obtained clearly demonstrate the effectiveness and efficiency of hBOA when applied to these two problems. Here hBOA is tested on a large real-life WDS design problem to further assess the effectiveness and especially the efficiency of the hBOA methodology. It is concluded that the ability of an evolutionary algorithm to learn the structure of a problem can provide significant improvements over simple genetic algorithms in terms ofthe quality of solutions, but that for particularly large networks the Bayesian network may not be the most appropriate methodology for learning the structure.