Nowadays, postal networks, air traortation systems and tourism companies use Hub Maximal Covering Model in order to benefit its economies of scale and optimal flow covering between origins and destinations. But, some hub nodes get congested specially in peak hours and this lead to improper covering and consequently an inefficient hub-and-spoke network. Common Hub Maximal Covering Models could not prevent this undesirable event. These models suppose that hub nodes always are available and there are an infinite number of servers for each hub. In order to handle this problem, this study considers hub nodes as an M/M/C queuing system, individually. Then, three Maximum Expected Hub Covering Location Models are presented, in which busy probability of hub nodes is considered. In the first model, busy probabilities of the servers are considered equal and the number of servers in different hubs is the same. In order to strengthen the model, the second model is presented which supposes busy probabilities and the number of servers in different hubs is not the same and given. This model uses the current number of servers in the hubs and might not be optimal to service the network. Finally, the third model is presented which determines the optimal number of servers considering various busy probabilities for servers in different hubs. Since the proposed models are NP-hard two Metaheuristic algorithms based on Genetic Algorithm and Tabu Search are developed. Computational results on the US domestic air traortation network in 2011 peak hour show that, in addition to covering radius, busy or free probability of servers has an important effect on locating hubs and allocating non-hubs to hubs. Also, computational results on the US domestic air traortation network show the proposed algorithms are efficient and can give near optimal solutions in a short time.