In the most past hub location problem (HLP) studies, failure probability is limited to a constant value; while in reality, it depends on various factors including natural disasters such as floods and earthquakes, commodity flow, institutional elements such as strikes, and etc. This paper looks into the problem of hub network design through a new model. The problem is to maximize the network reliability and to minimize the total routing cost simultaneously. In order to provide a more realistic model, the hub's failure probabilities were considered as a function of the hub's incoming traffic. Reserve hub elements are also taken into account in the model with the aim of increasing network reliability. To solve the model, a non-dominated sorting genetic algorithm (NSGA-II) is presented. The parameters of the algorithm are tuned using Taguchi method. The proposed solution is then carried out in a case study where the results confirm the acceptable performance of the model and also sensitivity analysis is performed in order to describe the effects of critical parameters into objective functions.