Synchrophasor-Based State Estimation for Microgrid Protection Dmitry Rimorov * , Yves Brissette † , Innocent Kamwa † and G´ eza Jo´ os * * School of Electrical and Computer Engineering McGill University, Montreal, Canada Emails: dmitry.rimorov@mail.mcgill.ca, geza.joos@mcgill.ca † Hydro-Quebec Research Institute, Varennes, Canada Emails: yves.brissette@ireq.ca, innocent.kamwa@ireq.ca Abstract—The challenge of microgrid protection is associated with many factors, such as frequently changing conditions and topology, lack of established power flow direction, presence of inverter-interfaced generation, among others. To overcome the challenges faced by the conventional protection, the paper demonstrates the application of a synchrophasor-based state estimation algorithm for protection of microgrids. The approach is tested on a model of an existing microgrid/test feeder. The results show the inherent flexibility of the approach towards changing conditions of microgrid operation. It requires minimum coordination that ensures quick detection of various types of faults in different locations. I. I NTRODUCTION Protection of active distribution grids and Microgrids (μG) is a challenging task for multiple reasons. Due to their op- erational flexibility, such systems are expected to frequently change topology and operating conditions. Therefore, dom- inant power flow patterns are hard to establish, which is a challenge for coordination of directional protection. Issues like nuisance tripping and protection blinding [1] are expected to occur more often. Furthermore, large share of distributed generation, in particular inverter-interfaced sources alters cur- rent contribution levels, making the utilization of conven- tional over-current or distance protection relays problematic; moreover, the control mode of inverters may affect relay performance [2]. There is a growing consensus that protection schemes must become adaptive in some way. Proposed approaches are viewed to be equally applicable to active distribution systems and μGs. One popular approach is to ensure that the settings of the conventional relays are changed in response to changing system conditions. These settings can be calculated in real- time, for instance by using a fault analysis simulation tool [3], or by utilizing an optimization approach [1]. Alternatively, relay setting groups can be pre-determined in advance for different conditions of the network [3]. Another direction for protection approaches that can be ap- plied in μGs explores the idea of ”setting-less“ protection [4] based on state estimation (SE). The main advantage of such ap- proach is its inherent flexibility similar to differential or model- based protection. Dynamic state estimation algorithms and SE- based indices (particularly, chi-square index) are proposed to be used to check relative health and detect abnormalities within a protected element [5], [6]. Growing availability of high- precision PMUs is of particular interest for such applications [7]. In fact, protection functionality can be viewed as one of the PMU enabled features in μGs and distribution systems, along with state estimation for the purpose of monitoring, control and optimized operation. Feasibility of applying PMU- based state estimation for active distribution protection was demonstrated by Pignati et al. [8]. Rather than utilizing a chi-square index as fault indicator, several parallel augmented state estimators were used, each of which modeled a fault on a particular line. The paper also demonstrated viability of using PMU-based protection, as it has been shown to provide quick response and adequate precision, even with substantial measurement noise. The μG protection scheme proposed in this paper applies static linear state estimation based on phasor measurements obtained from an actual PMU algorithm. Instead of assuming a PMU at each bus, the approach utilizes limited measurements and considers various formulations of the SE problem that can reliably detect faults. The approach is tested for multiple scenarios that include various fault locations and fault types. The model of an existing μG was used. The rest of the paper is organized as follows: Section II describes the utilized approach, including the discussion on the benchmark and its modeling, PMU algorithm and SE formulation; Section III shows the results on the performance of the protection scheme; Section IV discusses the results and acknowledges the limitations of the utilized approach, outlining the future work; Section V concludes the paper. II. PROPOSED APPROACH A. IREQ Test System and Its Modelling The model of the μG used in the paper is based on an existing test line – the IREQ test feeder (Fig. 1). It is a small 25 kV network that can operate either in islanded mode or connected to a 120 kV grid, and thus is qualified as a μG. It consists of a battery storage device (250 kVA), an inverter- interface generation source (250 kVA) capable of following MPPT of a renewable source, a synchronous generator (400 kVA) and a controllable load. They are connected to dedicated feeders of approximately 150 m each (Lines 2-3 and 2-4) through 500 kVA transformers. Line 1-2 is about 200 m long. 978-1-5386-7703-2/18/$31.00 ©2018 IEEE Authorized licensed use limited to: BIBLIOTHEQUE DE L'UNIVERSITE LAVAL. Downloaded on October 02,2022 at 23:22:03 UTC from IEEE Xplore. Restrictions apply.