Water distribution network (WDN) is a typical real-world complex network of major infrastructure that plays an important role in human's daily life. In this paper, we explore the formation of isolated communities in WDN based on complex network theory. A graph-algebraic model is proposed to effectively detect the potential communities due to pipeline failures. This model can properly illustrate the connectivity and evolution of WDN during different stages of contingency events, and identify the emerging isolated communities through spectral analysis on Laplacian matrix. A case study on a practical urban WDN in China is conducted, and the consistency between the simulation results and the historical data are reported to showcase the feasibility and effectiveness of the proposed model. Nowadays, the quality of human's life is significantly dependent on a reliable water supply system, which is one of the most critical infrastructures in modern society. With fast development of metropolitans, the structural and operational complexities of modern cities are increasingly dependent on the efficacy of Water distribution network (WDN). Quite often, any partial system failure or malfunction in WDN are greatly amplified through the whole network and thereby causing unexpected cascading contingencies, which potentially could split the WDN into isolated communities. The serious subsequent consequences due to system split can involve immediate suspensions of water supply in multiple areas. To some extent, traditional technologies are insufficient to study complicated behaviors of WDN, especially under contingency conditions. The model being proposed based on the view point of network structure will serve to provide a new solution of isolated community detection in the WDN.