In this paper, an optimization scheme based on the Corridor Method (CM) matheuristic is proposed for power distribution network configuration with the objective of client balancing between feeders. The problem is formulated as a mixed integer optimization problem (MIP) referring to a variant of the constrained spanning forest problem on a network. The problem is classified as NP-complete in its decision version, and the proposed matheuristic optimization approach is aimed to find high quality solutions in a short computational time on large real instances. According to the characteristics of the power distribution networks, suitable modeling solutions are adopted to retain a radial structure, satisfy power flow and voltage constraints, and reduce the reconfiguration efforts requiring a limited number of maneuvers on the electrical network. Computational experiments on different instances related to a real network have been conducted. The results demonstrated the soundness and the effectiveness of the proposed solution with respect to a baseline represented by a MIP formulation solved by a state-of-the-art commercial solver: the scalability towards large-size cases has been also considered. These results show that the proposed configuration method can effectively and efficiently solve the considered problem giving adequate support to the decision-makers.