Optimizing chlorination in distribution water networks using multiobjective genetic algorithms

The aim of this work is the development of multiobjective models to optimize medium and short-term chlorination management in distribution water networks . Decision variables are the number, locations and scheduling of chlorine booster stations.

The problem formulation considers a healthy objective, expressed by the minimization of the deviation sum squares of free chlorine rates on the network, and two economic objectives, expressed by the minimization of the injected chlorine mass and the minimization of the number of used chlorine booster stations.

The chlorination optimization has been designed in two stages: The first is the determination of the number and the identification of booster stations locations in the distribution water network (DWN): Medium-term chlorination management. The scheduling optimization of known chlorination stations in the network constitutes the second stage. It is the short-term chlorination management. Thus, two elitist multiobjective genetic algorithms (MOGA), using the Pareto optimality concept, have been developed to resolve the formulated optimization models.

The elaborated methodology has been tested on a bibliographic water network, given with the EPANET software, and on a real water network. Large Pareto front have been calculated for each chlorination management term. They express the relations between the healthy chlorination objective and each of the economic ones. The optimal calculated solutions, for the bibliographic network, have been more efficient in terms of uniformity and of chlorine mass used, than those proposed by previous research works. The chlorination optimization of the real network lead to considerable improvement in the free chlorine rates spatio-temporal uniformity, in comparison with the initial situations. This improvement is essentially the result of the use of chlorination booster stations in the network. In addition, significant reduction of the injected chlorine masses has been obtained.

Multiobjective chlorination optimization by MOGA appeared to be an efficient tool for the decision-makers to get a large choice of optimal solutions. According to priorities, an optimal solution can be chosen among those in the Pareto front. This approach doesn’t need to redo the same calculations to every modification of the management priorities.