Cost Savings on Large Water Distribution Systems: Design through Genetic Algorithm Optimization

The optimal design of water distribution systems for expansion, rehabilitation, reinforcement or rezoning is a complex process requiring systematic examination and evaluation of the design alternatives. The issues that make these design problems difficult to solve are presented in this paper, and the case for formal optimization methods is made. The reasons for adopting the genetic algorithm (GA) as the search technique are explained, and an outline of the GA methodology is given. Design variables include pipe locations and diameters, valve status, pumps and reservoirs. The inclusion of pumps into the optimization process requires that both the design and the operation of the network should be considered. Provision of new or expansion of existing service reservoirs is also described. Including service reservoir storage in the optimization requires simulation of the filling and emptying of the reservoirs through the daily cycle of demands, in addition to analysis of the instantaneous peak and emergency flows. A new methodology is used to increase the efficiency of the GA and to include the reservoir balancing in the problem formulation. Finally, the problem of rezoning an existing system for leakage control is presented using a case study. Substantial savings on manually produced solutions are demonstrated. These savings justify the use of genetic algorithms for these types of problem, and indicate that most other manually designed schemes may be costing far more than is necessary.