Hub disruption result in substantially higher transportation cost and customer dissatisfaction. In this study, first a mathematical model to design hub-and-spoke networks under hub failure is presented. For a fast and inexpensive recovery, the proposed model constructs networks in which every single demand point will have a backup hub to be served from in case of disruption. The problem is formulated as a mixed integer quadratic program in a way that could be linearized without significantly increasing the number of variables. To further ease the model’ computational burden, indicator constraints are employed in the linearized model. The resulting formulation produced optimal solutions for small and some medium size instances. To tackle large problems, three efficient particle swarm optimisation-based metaheuristics which incorporate efficient solution representation, short-term memory and special crossover operator are proposed. We present the results for two scenarios relating to high and low probabilities of hub failures and provide managerial insight. The computational results, using problem instances with various sizes taken from CAB and TR datasets, confirm the effectiveness and efficiency of the proposed problem formulation and our new solution techniques.