In recent years, the technology of micro-grids (µGs) has gained a lot of interest. µGs utilize distributed generators (DGs) to locally meet the needs of consumers. This decentralized nature increases the efficiency of the system in addition to improving the electrical service reliability. In this paper, coalitional game theory is used to study the process of local power exchange among a set of µGs. A coalition forming (CF) algorithm which depends on the topology of the network is proposed. The presented algorithm is scalable i.e. can capture a substantial number of µGs which makes it applicable in real systems. Besides the macro-station (MS), each formed coalition comprises a set of µGs with a lack of power to purchase and a set of µGs that have an excess of power to sell. Within each coalition, quadratic programming (QP) is used to organize the transition of power among µGs and MS so as to reduce the power losses. To illustrate the significant impact of the presented procedure, a full numerical example is introduced. The analysis and simulation show that the proposed algorithm results in a decrease in the average power losses per bus, reaching up to 37.6% enhancement compared with the non-cooperative situation.