Survivable multi-commodity network flow design: case of node capacities and arc failure

Abstract

This paper is focused on the design of a survivable network with node capacities and flow restrictions. The goal is to design a survivable network at minimum cost so that feasible flows exist at the time of multiple simultaneous failure arcs (failure scenario). To solve this problem Benders' decomposition (BD) was initially proposed and, then, a new strategy was presented to obtain specific failure scenarios. We computationally demonstrated that BD using this strategy could obtain the optimal solution faster. This strategy reduced iterations more greatly than the BD approach. Using this strategy, the length CPU time required for solving the problem was reduced by 30% on average.