In many power networks, power supplies are usually set to be in a redundant configuration. The total power demands in the networks may be far less than that of the capacity of the power supplies. After the power demands fulfilled an energy saving power generation dispatch becomes possible, and it plays an important role in saving the excess power in the networks. In this paper, we studied the energy saving problem in distributed power systems with tree configurations. We formulated the problem as a cooperative game among bus agents in a decentralized way. The Nash Bargaining Solution (NBS) provides a Pareto optimal and our energy saving algorithm is fair to all participated agents. Finally, we applied the algorithm to different scaled bus-oriented microgrids and the simulation results showed that the feasibility and performance are promising.