Method for optimal tearing of electrical networks

Abstract A method for optimal tearing of electrical networks to be analysed by diakoptics is presented. Starting from the completely disconnected network, the nodes are aggregated in areas by dynamic programming. At each step, nodes are displaced in order to minimize the requirement of the total core memory of the network impedance or admittance matrices. The choice of nodes, which are the seeds of aggregation for the areas, is made with regard to the connectivity properties as evaluated by the nodal proximity matrix expressed in the form P = P ′ + e P 2 . The test results for the AEP 14, 30, 57 and IEEE 118 bus systems are also presented.