Flow-Sequential Sector-Specific Lumped Algorithm for Water Network Calibration

Abstract

As part of the Battle of the Water Calibration Networks (BWCN) an algorithm termed the Flowsequential Sector-specific Lumped Algorithm (FSL) is employed to determine a set of roughness values for each pipe and the user demands at every junction of the system used for the contest. The FSL algorithm for roughness determination lumps all pipes in each pressure zone (similar to given district metering areas) together and alters the roughness values proportionately as a whole. Fine tuning with the pipes adjacent to the known pressure nodes is the final step. This is done iteratively from the furthest upstream pressure zone to the furthest downstream pressure zone and repeated until a satisfactory agreement between predicted and measured value is attained. Age-dependent roughness values are initially assigned according to literature values. The network in the demand calibration phase of the study was handled as a whole. Demand values based on the monthly demands were provided, and a common residential diurnal pattern found in the literature was assigned as an initial value for the extended period simulation. The demand pattern was adjusted for select hours to calibrate the network to match the SCADA data. This was followed by the fine tuning of the demand pattern values. It was found that the calibration process resulted in mean absolute relative errors in pressure of approximately 0.1 when adjusting roughness values for calibration with fire flow tests and 0.21 for discharge and tank level values in the adjustment of demand values.