A recursive algorithm for reliability assessment in water distribution networks with applications of parallel programming techniques

Project models the reliability of an urban water distribution network.. Reliability is one of the fundamental considerations in the design of urban water distribution networks. The reliability of a network can be modelled by the probability of the connectedness of a stochastic graph. The enumeration of a set of cuts of the graph, and the calculation of the disjoint probability products of the cuts, are two fundamental steps in the network reliability assessment. An improved algorithm for the enumeration of all the minimal cutsets of a graph is presented. Based on this, a recursive algorithm for the enumeration of all Buzacott cuts (a particular set of ordered cuts) of a graph has been developed. The final algorithm presented in this thesis incorporates the enumeration of Buzacott cuts and the calculation of the disjoint probability products of the cuts to obtain the network reliability. As a result, it is tightly coupled, and very efficient. Experimental results show that this algorithm has a higher efficiency than other reported methods. The parallelism existing in the reliability assessment is investigated. The final algorithm has been implemented in a concurrent computer program. The effectiveness of parallel programming techniques in reducing the computing time required by the reliability assessment has also been discussed.