This paper addresses the problem of fault identification in meshed HVDC grids once an abnormal behavior has been detected. A parametric single-ended fault identification algorithm is proposed. The method is able to determine whether the line monitored by a relay is faulty or not using a very short observation window. When a fault is suspected, the proposed algorithm estimates the fault distance and impedance using a parametric model describing the voltage and current evolution just after the fault occurrence. This model combines phenomenological (knowledge-based) and behavioral (black-box) parts to represent the fault propagation and to account for ground effects and various losses. The identification of the faulty line is then based on the size of the confidence region of the obtained estimate. The performance of the algorithm for a three-node meshed grid is studied using Electro-Magnetic Transient (EMT) simulations. On the considered grid model, the current and voltage need to be observed during less than 0.2 ms to get a sufficiently accurate estimate of the fault characteristics and identify consistently the faulty line.
Paul Verrax, Alberto Bertinato, Michel Kieffer, Bertrand Raison
Published in Electric Power Systems Research, Volume 185, August 2020, 106387