2 nd International CCWI / WDSA Joint Conference, Beijing, China – November 12-15, 2020 Battle of the Leakage Detection and Isolation Methods: An Energy Method Analysis using Genetic Algorithms Juan Saldarriaga 1 , Laura Solarte 2 , Camilo Salcedo 3 , Carlos Montes 4 , Laura Martínez 5 , María González 6 , María Cuello 7 , Andrés Ariza 8 , Camilo Galindo 9 , Néstor Ortiz 10 , Cristian Gómez 11 & Sergio Vanegas 12 Universidad de los Andes, Civil and Environmental Engineering Department and Water Supply and Sewerage Systems Research Center (CIACUA), Carrera 1 Este # 19A - 40, Bogotá, Colombia 1 jsaldarr@uniandes.edu.co, 2 lm.solarte@uniandes.edu.co, 3 ca.salcedo959@uniandes.edu.co, 4 cd.montes1256@uniandes.edu.co, 5 ls.martinez10@uniandes.edu.co, 6 ma.gonzalezm1@uniandes.edu.co, 7 mm.cuello547@uniandes.edu.co, 8 ad.ariza@uniandes.edu.co, 9 c.galindo@uniandes.edu.co, 10 nv.ortiz@uniandes.edu.co, 11 cc.gomez@uniandes.edu.co & 12 sm.vanegas@uniandes.edu.co Keywords: Leakages, emitters, Genetic Algorithms, EPANET, water networks. 1. Context: Problem approach Leakages in water networks are a common issue in urban water supply infrastructure. These leakages can occur in pipe junctions and sections and in pipes or accessories that are part of storage/compensation tanks. Usually, leakages can be caused by ground movement or settling due to traffic or poor-quality materials. In addition, pressure fluctuations in the network, pressure transitions, or excessive pressures are a common cause of failures leading to leakages. When the behavior of leaks is analyzed, it can be seen that they are dependent of the demand, something that varies along the day, because during low demand periods, network pressures are higher, and this results in larger leaks discharges. This behavior can be noted in Figure 1 and Figure 2, the first shows fluctuations of flow along a 24h period for a node before a leak event, after it and the case after leak detection using a hydraulic model. Figure 2 shows how a higher pressure generates a higher demand and leakage. The equation from an emitter can be used to simulate leakage flowrate in a particular node in a water distribution network (WDN) (Wu, et al., 2010). Figure 1.Comparison between flowrate before and after a leak detection throughout a day. Wu et al, 2010. Figure 2.Relation between pressure, demand and leakage in node i. Gupta et al, 2016. The methodology used to address the objective of this battle requires a network calibrated hydraulic modeling, which requires detailed information of topological, topographical and hydraulic variables. When calibrating the model, possible leaks are simulated in the nodes using emitters, where the leak flowrate is assumed to be a function of the pressure at that point as indicated in the Equation 1.   =      [Equation 1]