Classic data envelopment analysis (DEA) models treat the decision making unit (DMU) as a “Black Box” which without taking into consideration the internal processes and their efficiencies, enters the Input and gives the Output and thus calculates the efficiency DMU as a whole. Whereas, inefficiency of any DMU is mainly the result of inefficiency of internal parts of The DMU. Thus, Network DEA method was presented for the evaluation of the efficiency of similar DMUs with respect to their internal processes. Lewis and Sexton (2004) have presented one of the most useful methods of the Network DEA. In spite of this, their method has its difficulties, which includes the use of an inappropriate efficiency index, its inability to analyze different cycles in its internal processes, and not allocating an optimum intermediate output for proceeding sections. In this thesis, at first we attempted to solve these problems and then we defined an inter-DMU input-output variable to simulate the existing flow of interactions between different DMU’s. Finally taking into account our expansions, we presented a new advanced method that calculated the efficiency of different DMUs with complicated structures.