This paper presents a statistical voltage sensitivity analysis applied to radial distribution networks with distributed generation. The search focuses on finding the best nodes of a radial distribution network, where the power injections improve voltages magnitudes, while the constraints are satisfied. Two algorithms were used to find the best location, particle swarm optimization (PSO) and simulated annealing (SA). We changed the power injection from 10% to 60% of the total power load, by using solar and wind generators as scenarios to find the candidate nodes for the installation of new electrical sources.