In this thesis, we survey a p-hub median that the aim is determining of the hub location and allocate non-hub location with hub location so that the traortation cost will be minimum. we introduce the Dynamic Uncapacitated Hub Location Problem (DUHLP) which consists in minimizing the total cost over a finite time planning horizon while ensuring that at each single period all demand is fully routed through the network. There are two assumptions underlying in this model. The first is that all flows have to be consolidated by hubs. Thus, the paths between O/D pairs must include at least one hub node. The second is that it is possible to fully interconnect hubs with more effective, higher volume pathways that allow a discount factor to be applied to the traortation cost of the flows between any pair of hubs. The proposed model has been considered as multi-period with discrete demands. Periods have been considered as discrete and with constricted. Hub location is assumed fixed. Also for hub location, after end of the period, recovery gain is considered. The objective function value count as net present worth. In this thesis it has been tried to present the model which the hub network to be as incomplete network based on traortation cost so that traortation cost will be minimum. Namely, there is not necessarily direct contact between two hub location. Because of the difficulty to solve the question and NP-hard nature and also in accordance with considered assumes , it has been tried to use fewer constraints and variables. Also because to reduce variables, we propose a reduction principal. The model is formulated as a binary non-linear programming. In this model we investigate a dynamic, time-dependent, multi commodity where we establish new facilities over a given time horizon. Despite the difficulty of solving the problem and the nonlinear mathematical model of the third degree in the objective function, showed that solving the model common methods of optimization is not possible in reasonable time. This model is formulated as a mixed binary non-linear problem that we propose two solution procedure. First procedure is based on branch and bound algorithms and second procedure is based on genetic algorithms. In first procedure for calculating lower bound we propose a lagrange approach which relaxes some constraints. The resulting model has been divided two subproblem and each subproblem has been solved. Briefly, the Lagrange function exploits the structure of the problem and can be decomposed into smaller subproblems which can be solved efficiently and we use of a sub gradient optimization method to improve the lower bound. Then, we proposed a heuristic procedure to solve this model with large size. Population in genetic algorithm proposed is included the original population and the subpopulation. The chromosomes is Used in the subpopulation, been considered as a matrix. A sample of 80 problem generated based on CAB data set and this model was solved Using methods presented and results has been studied. Results of sample problem compare and the performance of the proposed genetic algorithm is shown. Also according to the results obtained, the branch and bound algorithm proposed is not suitable for large size problems