Optimum Design Of Water Networks Using A Micro Software Based Genetic Algorithms Multiobjective

Edwin Pino, Franz Condori, Luis Alfaro, Angely Valle

Tuesday 30 june 2015

12:39 - 12:42h at Europe 1 & 2 (level 0)

Themes: (T) Water engineering, (ST) Computational methods, Poster pitches

Parallel session: Poster pitches: 5E. Engineering - Computional

The water distribution networks make a system that seeks to bring water to each user and according to their use. The infrastructure required to accomplish this objective requires a significant financial investment, thus providing the optimal design of these networks is very important. Not only the economic aspect is paramount, besides him there are aspects such as the hydraulic behavior referred to pressures and minimum speeds, the availability of pipelines, water quality, the distribution of demand, network reliability, system operation, complicated making a comprehensive analysis that may lead to an optimal design. The solution to the optimization problem of a water network problem is a NP-HARD combinatorial called, which means it is not possible to use a deterministic method to solve it, but the solution requires special methodologies that making use a reasonable computational processing time allows obtain a configuration that meets the constraints diameters (such as velocity and pressure) and obtain acceptable values in the objectives sought (such as the search for lower cost and higher reliability). The use of genetic algorithms can solve this kind of NP-HARD combinatorial problems, obtaining the optimization of a water network. For optimal set of solutions, was developed a computer program which is based on the use of a Micro Multi-objective Genetic Algorithm, proposed by Toscano and Coello (2001), adapted to the application of water distribution networks, which was developed seeks to fulfill in addition to economic constraints, also meet under Peruvian law, pressure and speed limits. The computer program has been developed in the MATLAB programming language. Finally proceeded to check the effectiveness of the algorithm proposed in the Hanoi network (Fujiwara and Khang). This is a network that has a single source; with 3 basic circuits, 31 nodes, a reservoir and 34 tubes. The nodes are located at the same elevation and are not considered minor losses in pipes.