Prioritizing Pipe Replacement in a Water Distribution System Using a Seismic-Based Artificial Neural Network Model

Abstract

This work focused on developing an approach for prioritizing the order of pipe replacement in a water distribution system (WDS) using a seismic-based artificial neural network (ANN). The qualified earthquake data obtained from the Taiwan Water Corporation Leakage Repair Management System (TWC-LRMS) were classified to build the model that was analyzed by both backward propagation network (BPN) and radial basis function network (RBFN). Pipe diameter, pipe material, and the number of monthly magnitude-3+ earthquakes provide the input parameters of the seismic-based ANN model for anticipating the priority of pipe replacement. The WDS of Yilan County, which frequently suffers from earthquakes in northeastern Taiwan, was used as the object of the case study. A comparison of the accuracy and reliability of the prediction model between BPN and RBFN demonstrated that RBFN outperformed BPN. The seismic-based ANN model developed in this work is streamlined for establishing a priority project of pipe replacement. The number of breaks predicted by the ANN model was close to the observed data. Furthermore, ANN has qualified as an effective technology for developing feasible pipe replacement priority in the domain of water leakage management.