A New Adaptive Underfrequency Load Shedding Scheme to Improve Frequency Stability in Electric Power System Fethi Boussadia * , Saad Belkhiat Electrical Engineering Department, Faculty of Technology, Ferhat Abbas Setif 1 University, Sétif 19000, Algeria Corresponding Author Email: fboussadia@univ-setif.dz https://doi.org/10.18280/jesa.540208 ABSTRACT Received: 7 December 2020 Accepted: 25 March 2021 This paper proposes a new adaptive underfrequency load shedding scheme (UFLS) to avoid frequency instability in electrical power system during abnormal wide area disturbances. The developed scheme is based on online monitoring of the distance relay zone 3 decisions of some tie lines using (WAMS) and SCADA/EMS systems, to check rapidly and reliably the uncontrolled islanding conditions and, permit an automatic load shedding action to maintain frequency stability of power system. Simulation results on 400 kV Turkish transmission systems demonstrated the effectiveness of the proposed scheme compared to current frequency load shedding schemes which they cannot consider all possible circumstances because of their limited access to the power network data. Simulation results clearly indicate that large disturbances in power systems can be avoided and their propagation can also be stopped using the proposed scheme. Keywords: frequency control, load shedding scheme, real time measurement, generation shedding, emergency conditions 1. INTRODUCTION During later decades, many major blackouts have been experienced in different countries of the world. On July 31, 2012, the most severe blackout in history of the word was occurred in Indian Electricity Grids. This blackout affected more than 620 million people and about 1.4 GW power was lost. During the system disturbance the Indian power system was separated into three areas (Northern-Eastern, Eastern and Western areas). The enormous unbalance between load and supply in each area caused a significant frequency deviate. This situation led to a total power system blackout in around 20s [1]. On September, 28 th , 2003, the tripping of some heavily loaded 400 kV transmission lines at the corridor between Italy and its neighboring systems, led to the disconnection of the Italian power system from the UCTE power system. This separation caused a loss of 6651 MW power import and the frequency rapidly dropped. The operation of the primary frequency control and UFLS scheme could not stop the frequency decrease. Finally, the frequency dropped further and caused a total blackout [2, 3]. Also, In Turkey on March 31, 2015 a cascading failure of some tie lines caused interruption of all customers connected to the network of the country with a total of 33450 MW lost [4]. During the system disturbance the Eastern and Western Turkish subsystems were separated. The huge imbalance of about 4700 MW that have experienced the Eastern and Western subsystem caused the collapse of the Network in around 10 sec. Indeed, the causes of the instability incidents mentioned above are multiple and one of them is directly related to the inability of load shedding scheme to bring the frequency back to a safer level in the formed islands after the system split [5]. In fact, UFLS scheme being used to mitigate a rapid decrease in frequency due to unbalance between production and consumption in power system [6]. However, current UFLS schemes used by most energy utilities around the world are based on static models in which the thresholds and the load to be shed are based only on frequency measurements [7]. So, the amounts of load that be shed by these schemes are fixe irrespective of the magnitude of disturbance [8]. This may lead to under-shedding and over shedding problems and can further deteriorate the frequency stability of the electrical network [9-11]. In order to improve the operation of these conventional UFLS schemes during instability conditions, different techniques are proposed in the literature. UFLS Schemes based on the measurement of the rate of the frequency change (df / dt) have been designed to avoid the drawbacks of UFLS scheme which was based only on frequency amplitude measurements [12, 13]. Thus, the power deficit is calculated immediately after the disturbance and the load to be shed will be optimized according to variations of electrical system parameters as well as the dynamics of the disturbance. However, during various transient events the measurements of the frequency and its derivative can present large errors due to rapid variation of the conditions of the electrical network. In this case, the UFLS could not shed accurate amount of load and the power deficit estimation will be affected [14, 15]. The use of a hybrid load shedding scheme which takes into account both frequency and voltage variations has been studied in reference [16]. So, this scheme allowed shedding appropriate amount of load using both voltage and frequency load shedding relays. Different artificial intelligence-based schemes have been also employed to accurately estimate the power deficit during frequency instability conditions [17-21]. Centralized load shedding schemes have been suggested to execute a global load shedding decision using modern communication devices as reported in the references [22-24]. However, in practice, the lack of information about the amount of load that has been shed by frequency load shedding relays would greatly affect the reliability of these schemes. The mentioned weakness can be overcome by the use of new schemes that based on Wide Area Measurement System Journal Européen des Systèmes Automatisés Vol. 54, No. 2, April, 2021, pp. 263-271 Journal homepage: http://iieta.org/journals/jesa 263