Case Study: Pipe Network Model Calibration Issues

Abstract

This paper illustrates that it is possible to calibrate accurately a mathematical model of a complicated water distribution system (271 nodes, 307 pipes, 9 pressure reducing valves) if adequate and accurate data are collected. It is essential that the model be calibrated for a range of water use rates and patterns as otherwise it is possible for the model to appear to be calibrated even though it contains significant errors. For example, in this system it was possible to make the model appear calibrated for C‐factors that varied by a factor of 3 provided that sufficient errors were made on other input parameters. The key to collecting adequate data is that the head at boundary nodes (pumps, tanks, pressure reducing valves) be known when pressure readings are made in the system.