The seismic reliability of a water supply network is the service level of the network underpossible seismic intensities， which is mostly estimated using Monte Carlo simulations to generate alarge number of seismic damage samples， and this method will bring great workload and ultra-hightime cost for seismic reliability assessment of large and complex networks. In order to overcome theseproblems， not only a multi-burst pipe pressure drop calculation model is proposed， but also a scenarioreduction method is introduced to improve the calculation efficiency. Firstly， the multi-burst pipe pres‐sure drop model is used to calculate the nodal water pressure under the seismic damage scenario. Sec‐ondly， the scenario merging and classification of nodal water pressure in seismic damage is performedusing the cluster averaging method in system clustering. Then， k-medoids clustering is used to selectthe center point of each cluster as a typical scenario. Finally， the node reliability index and system reli‐ability index are determined by the typical scenarios. The proposed algorithm is applied to a practical pipe network and compared with the results of traditional Monte Carlo simulation. The results showthat the multi-burst pipe pressure drop model based on linear estimation is reasonable； the reliabilityassessment based on the scenario reduction method is feasible； and increasing the number of typicalscenarios can reduce the assessment error. Therefore， the proposed algorithm can ensure the accuracyof the results while improving the computational efficiency.
