Simultaneous Calibration of Leakages, Demands and Losses from Measurements. Application to the Guayaquil Network (Ecuador)

Abstract Hydraulic models of water supply networks are widely used by utility companies to assist in decision making. The reliability of the model strongly depends on the quality of its calibration, that is, the correspondence between the measure variables and the calculated ones. When dealing the model is static, the calibration is easy to reach for average values of the variables. On the contrary, dynamic models need to spatially allocate demands and distribute them among the nodes of the network, which complicate the problem. The paper proposes a methodology for preliminary calibration of hydraulic models based on advanced calibration techniques. This methodology is applied to models with both pressure-dependent and independent demands. Pressure-dependent demands are related to leakages and are spatially distributed according to the length of the pipes and volumetric efficiency of the district metering area (DMA) being considered. In order to model leakages, the Germanopoulos model has been chosen. Thus, leakage flow is a function of the length of the pipes and the pressure along them. The equation to quantify the volume of leakage is a modification of the orifice equation. From this leakage model, a calibration process is proposed. It consists in three steps. First, a global leakage coefficient is calculated in order to satisfy daily mass balance of produced unaccounted and consumed water. In the second step, a time demand pattern is calculated. After these two steps a preliminary model is obtained. Finally, a conventional calibration process is done using discrepancies between pressure measurements and model result to adjust both roughness and minor losses coefficients. In order to validate this calibration methodology, a case study was used in Guayaquil (Ecuador) in which three DMA were studied. The results showed that the method converges very quickly and is effective regardless the volumetric efficiency of the network.