Report Open Access

Battle of the Leakage Detection and Isolation Methods: An Energy Method Analysis using Genetic Algorithms

SALDARRIAGA, Juan; SOLARTE, Laura; SALCEDO, Camilo; MONTES, Carlos; MARTÍNEZ, Laura; GONZÁLEZ, María; CUELLO, María; ARIZA, Andrés; GALINDO, Camilo; ORTIZ, Néstor; GÓMEZ, Cristian; VANEGAS, Sergio

Solving the problem of this battle required the use of a calibrated hydraulic model for the network. Within the hydraulic modelling, the nodes represent possible water entrances (from tanks, for example) and exits (demands and/or leakages) in the distribution network. When calibrating the model, the leakages are simulated in the nodes through emitters, where the flowrate from leakage is assumed to be a function of the pressure at that point as previously indicated in the emitter. In the emitter equation, our methodology assumed a constant exponent of 0.5 and a variable coefficient to be determined with the GA.

The use of GA was proposed due to its wide use in nonlinear problems as identification of leaks in a Water Distribution Network WDN. Its potential relies on the optimization of a function through the combination and mutation of the data which allows the method to skip local optimal and approach a global optimal. The performance of the GA revolves around the function which varies according to the RMSE between the pressure signals calculated and recorded in the database. The performance of the GA is highly linked to the number of generations modelled because more alternatives can be considered, which generates a higher probability of achieving an optimal. Additionally, the genes that make up each individual are the topographic coordinates of the leaking node, the coefficient of variation, and the time at which the leakage occurs, in order to summarize the hydraulic behavior of leakages in the network. The steps to develop the proposed methodology are explained in detail in the extended abstract above.

Files (334.3 kB)
Name Size
334.3 kB Download
All versions This version
Views 157157
Downloads 133133
Data volume 44.5 MB44.5 MB
Unique views 144144
Unique downloads 117117


Cite as