1 Abstract--This paper proposes a new approach for optimal phasor measurement units placement for fault location on electric power distribution systems using Greedy Randomized Adaptive Search Procedure metaheuristic and Monte Carlo simulation. The optimized placement model herein proposed is a general methodology that can be used to place devices aiming to record the voltage sag magnitudes for any fault location algorithm that uses voltage information measured at a limited set of nodes along the feeder. An overhead, three-phase, three-wire, 13.8 kV, 134- node, real-life feeder model is used to evaluate the algorithm. Tests show that the results of the fault location methodology were improved thanks to the new optimized allocation of the meters pinpointed using this methodology. Index Terms— Optimal allocation, phasor measurement units (PMUs), fault location, Monte Carlo methods and optimization methods. I. NOMENCLATURE NM : number of meters; NB : number of nodes of the system; nf : number of faults applied; np : number of positions considered for sum of distances; dist MAX : max distance between faulted point and located one; dpf_pl k,i : distance between faulted point k and located point i; b i : node i with voltage sag greater than a preset value; Z : random number generated by normal distribution; μ : mean; σ : std. deviation; f(Z,μ,σ) : function of normal probability density; ε : random variable with normal distribution;; |CL| : size of candidate list; |RCL| : size of restricted candidate list  : control parameter of RCL size; R pg : single-line-to-ground resistance; R pp : resistance between phases; Bs : best solution; As : actual solution; iter : number of iterations; ITERmax : max number of iterations; V Pj [i] : voltage at phase j, node i; S nom : nominal apparent power cosφ : power factor This work was supported in part by the National Council for Scientific and Technological Development (CNPq) and in part by Foundation of Teaching, Research and Extension of Ilha Solteira (FEPISA). A. A. P. Bíscaro, R. A. F. Pereira and J. R. S. Mantovani are with UNESP – Univ Estadual Paulista, FEIS, Electrical Engineering Department, Ilha Solteira, São Paulo, 15385-000, Brazil (e-mails:aapbiscaro@hotmail.com; ddigo@yahoo.com; mant@dee.feis.unesp.br) II. INTRODUCTION ower systems are subject to several problems of either deterministic or stochastic nature, which can compromise the supply quality, resulting in losses for both electric utilities and customers. Among these problems, those related to voltage sags and fault location in electric power distribution systems can be emphasized. The focus of this paper is the optimal phasor measurement units (PMUs) placement for fault location. These devices are related to the analysis and control of voltage magnitude in the network and computer algorithms that use these voltage measurements to perform the fault location. The optimized allocation of such devices on the feeder and digital processing of the measures contribute directly to the efficiency, accuracy and reliability of fault location algorithms. In the literature, there are a few works related to the optimized allocation of phasor measurement units for fault location in electric power distribution systems. The proposals differ basically in methodology or optimization technique used. Reference [1] proposes a method based on fuzzy inference systems to make the allocation of fault indicator devices. The allocation is done considering only the main feeder. Reference [2] proposes a tabu search algorithm aiming to place the measurement devices. This algorithm explore the structure of historical data of the searches performed, using this information as a condition for the new movement in the search space. Reference [3] presents a comparison of various methods for optimal allocation of PMUs in order to make the estimation of network state. A graph method, based on rules, is initially used to solve the problem. Then, three metaheuristics are presented, namely: simulated annealing, genetic algorithm and tabu search. The mathematical model herein proposed in this paper for optimal allocation of such equipment in distribution feeders considers the fault location algorithm presented in [4] as an auxiliary tool to evaluate the objective function of the problem. The mathematical model seeks to define the position where the meters must be installed on the feeder, to establish the best cost/benefit relation of investments as well as to provide results with suitable quality and accuracy for any fault location algorithm that uses voltage sags information to locate faults. Optimal Phasor Measurement Units Placement for Fault Location on Overhead Electric Power Distribution Feeders A. A. P. Bíscaro, Student Member, IEEE, R. A. F. Pereira and J. R. S. Mantovani, Member, IEEE P