open access

https://hdl.handle.net/2115/100456

Data-Driven Reliability Assessment of Water Distribution Networks

英語

Uncertainty

2026-07-01

Proceedings of the 10th International Symposium on Reliability Engineering and Risk Management (ISRERM 2026)

Water distribution networks must deliver reliable drinking water service while operating under uncertain demand conditions. Previous operational optimization and reliability studies commonly represent demand uncertainty using parametric probability distributions or ellipsoidal uncertainty. This study investigates whether ellipsoidal uncertainty provides an appropriate representation of short-term operational variability in real water distribution systems. Five years of hourly demand data from a Canadian water distribution network serving approximately 100,000 residents are analyzed to characterize temporal demand structure and deviations from nominal demand profiles. Demand variability is examined, and the statistical properties of hourly demand deviations are evaluated to assess consistency with symmetric ellipsoidal uncertainty assumptions. Results show that demand variability exhibits clear seasonal structure, time-of-day dependence, and asymmetric deviation patterns that are not consistent with ellipsoidal uncertainty representations used in previous studies. These findings point to the need for incorporating empirically derived variability structures into water network reliability assessments.

会議発表論文

Version of Record