Abstarct: The sub-transmission substation expansion planning (SSEP) is a complicated problem and the aim of solving this problem is to determine optimal number, locations, capacities, and associated service area of all substations considering several technical constraints (such as voltage drop, loading capacity of substations, thermal limit of feeders, and radial network), at the possible minimum cost. In this thesis, two novel heuristic combination methods baxsed on clustering technique is presented to solve the SSEP problem in two steps: (1) Optimal substation placement, (2) Optimal substation allocation. In the first method, the k-means algorithm is developed by setting power demand of loads in calculating the mean of each cluster as a weight coefficient. This leads to place substations that are closer to big loads without any predetermined candidate locations. In the next step, by using dynamic programing (DP), optimal capacities and associated service area of substation with aim of minimizing investment and operational costs is determined. In this regard, a new connection cost including the cost of expansion capacity of substations is added to the costs of MV network. So that in the new approach, the loads are connected to that substations that no need to expand their capacities. Thus, the maximum existing network potential is used and the asset management to be considered. The second method presents a methodology solution to solve the SSEP for real large-scal network by the targeted decomposition of planning area into isolated small zones. So that the optimal substation placement problem is to be solved separately baxsed on the geographical information system (GIS) for each zone. To do so, the imperialist competition algorithm baxsed on k-means clustering technique (ICA-KCT) is employed. The discret problem of the best priority sequence of load connection is solved using discret ICA (DICA) to find out optimal associated service area of substations. Hence, the main ICA operators (assimilation and revolution) are utilized, so that an approximate prioritization heuristic method is used to apply the revolution operator process, in order to make DICA to be robust. Finally, the proposed method for solving the SSEP problem is applied to a typical test system and sub-transmission network of Mashhad. The findings of simulation verify the efficiency and robustness of proposed methods.