Many of network facility location models have predefined network structure. So designing the network and locating new facilities simultaneously should considered as an important subject area and there is an essential need to new models for optimizing these problems. Also, because of network failures in some cases, links and facilities can’t serve customers so addressing the network reliability and the congestion effects are important factors. To make the problem more functional, in this research the facility location network design problem is based and the congestion of link is added. Two meta heuristic algorithm, simulated annealing and variable neighborhood search, are used to solve this problem. To do so, a greedy heuristic algorithm is used to generate initial solution. SA uses a binary string to show nodes that contain facility and change the priority of facilities to create neighborhood. VNS create a stochastic order of nodes, then the number of facility change and supposed as a new neighborhood, when the best solution is defined a shaking happen and new order of nodes are made. The efficiency of algorithm is proved by numeral instances. Also the result of comparison of two algorithms in different real problems are examined.