A Spatio-Temporal exploratory analysis of pipe-failure incidents in the Water Distribution Network of Limassol, Cyprus

Abstract

This paper presents the first attempt to the use of Geographic Information Systems technology in the Water Distribution Networks of Cyprus. A variety of exploratory, statistical and visualization techniques were used in order to identify patterns in the degradation of the WDN of Limassol. The distribution of pipe-failure incidents was studied at the level of District Metered Areas (DMA) boundary classifications and at the location level using point-pattern analysis methodologies. A modified version of the DBSCAN clustering algorithm was used, in order to perform clustering using the Dijkstra network distances between the pipe-failure incidents in Water Mains. The number of pipe-failure incidents was found to increase rapidly during the years 2009 and 2010 and that this increase can be attributed to problems occurring in house connection polyethylene pipes of small diameters. The pipe-failure incident rate across the DMAs was not found to vary significantly. Point-pattern analysis reported strong spatial clustering between the pipe-failure incidents. The Network DBSCAN clustering approach was able to extract spatial clusters at different hierarchy levels and showed that dense clusters of water main incidents occur within the DMAs. This suggests that a localized indicator of degradation could provide important information for the sustainable management of Water Distribution Networks.