The fall of a conductor in a High Voltage (HV) network can have severe consequences. When the contact with the ground happens on the load side, this type of fault typically generates a very low fault current. Therefore, this situation may not activate the upstream protective systems in the network, so the fault remains unnoticed while a live conductor is in direct contact with its surroundings. We developed an algorithm to detect this issue to allow prompt action using data measured in the Low Voltage (LV) network. The purpose of this work is to assess whether the algorithm developed by Eneida, to detect and classify this type of fault, behaves as expected. When a phase is lost, meaning that the conductor was severed and is not in contact with any surface, a HV Phase Loss alarm should be raised. When the severed conductor is in contact with a surface, as simulated in these tests by being connected to the ground through a resistance, it should trigger a HV Line Down alarm. Currently, to pinpoint the faulty section in a HV network, several methods are available: travelling waves [2], [5], fault passage indicators [1], [10] and impedance methods [4]. Our method differs from these by using LV measurements and taking advantage of the existing LV monitoring infrastructure to detect, classify and pinpoint the fault. To assessourmethod’sperformance, wecarried out experiments at PNDC. In eachexperiment, one phase in the HV network was disconnected from the supply side, and connected to the ground via different resistors, thus throwing a controlled fault. The different fault resistances for each phase were as follows: Phase A: 80Ω, 90Ω, 120Ω, 140Ω, 150Ω, 180Ω Phase B: 20Ω, 30Ω, 60Ω Phase C: 110Ω, 120Ω, 150Ω, 200Ω, 210Ω, 240Ω The following metrics were used to assess our algorithm’s performance: M1: Percentage of events detected M2: Percentage of events correctly classified (HV Line Down / HV Phase Loss) In total, 15 experiments were performed. It was considered that, in order to validate the solution, the acceptance criteria should be met for, at least, 14 of the 15 tests (i.e. an error margin of a single test was considered). Therefore, the detection and classification of both eventsHV Phase Loss, followed by HV Line Down- must present an accuracy of, at least, 93.3% (acceptance threshold). In all the 15 experiments, we observed that both events- line disconnected from supply side, simulating a conductor breaking (HV Phase Loss), followed by a connection to the ground on the load side, simulating a conductor hitting the ground (HV Line Down)- were correctly detected, classified and located within the network. Therefore, as a result of these experiments, the solution under test is considered validated.