In a DC microgrid, several paralleled conversion systems are installed in distributed substations for transferring power from external grid to a DC microgrid. Droop control is used for the distributed load sharing among all the DC/DC converters. Considering the typical efficiency feature of power electronic converters, optimization method can be implemented in tertiary level for improving the overall system efficiency. However, optimization purposes usually require centralized communication, data acquisition and computation which might be either impractical or costly for dispersed systems. Accordingly, this paper proposes a dynamic consensus algorithm based distributed optimization method aiming at improving the system efficiency while offering higher expandability and flexibility when compared to centralized control. Hardware-in-the-loop (HIL) results are shown to demonstrate the effectiveness of the method.