Impact of Degraded Communication on Interdependent Power Systems: The Application of Grid Splitting

Di-An Tian, Giovanni Sansavini
2016 Electronics  
Communication is increasingly present for managing and controlling critical infrastructures strengthening their cyber interdependencies. In electric power systems, grid splitting is a topical communication-critical application. It amounts to separating a power system into islands in response to an impending instability, e.g., loss of generator synchronism due to a component fault, by appropriately disconnecting transmission lines and grouping synchronous generators. The successful application
more » ... grid splitting depends on the communication infrastructure to collect system-wide synchronized measurements and to relay the command to open line switches. Grid splitting may be ineffective if communication is degraded and its outcome may also depend on the system loading conditions. This paper investigates the effects of degraded communication and load variability on grid splitting. To this aim, a communication delay model is coupled with a transient electrical model and applied to the IEEE 39-Bus and the IEEE 118-Bus Test System. Case studies show that the loss of generator synchronism following a fault is mitigated by timely splitting the network into islands. On the other hand, the results show that communication delays and increased network flows can degrade the performance of grid splitting. The developed framework enables the identification of the requirements of the dedicated communication infrastructure for a successful grid-splitting procedure. Electronics 2016, 5, 49 2 of 25 assure reliable operations of the power grid by balancing demand and supply [3, 4] . Under these circumstances, disturbances, such as short circuits caused by accidental contact of transmission lines, e.g., with trees or cranes, lightning or strong winds causing the galloping of transmission lines, may lead to severe consequences on the operations of the electric power system, e.g., cascading outages experienced during the Northeast or Italian blackout in 2003 [5, 6] . In view of operations closer to stability limits, the use of ICT is expected to turn the current grid into a "smart grid" [7], making it self-healing and increasing its efficiency, reliability, security and quality of service [8] . On the transmission level, a wide area measurement system (WAMS) is being installed with phasor measurement units (PMUs) and phasor data concentrators (PDCs) at strategic locations in the grid [9] . A key feature is the availability of synchronized time tags for measurements with an accuracy of 1 µs through a global positioning system (GPS) receiver enabling the real-time monitoring, control and protection of the power system. An application that benefits from the WAMS is grid splitting [10], also referred to as controlled islanding, which relies on real-time system-wide measurements to enable the detection and recovery from failures in real time, i.e., by applying system topology changes. Grid splitting is a special protection scheme that separates a power system into synchronized islands in a controlled manner in response to an impending instability, i.e., generator rotation desynchronization triggered by a component fault. By appropriately disconnecting transmission lines, severe consequences, e.g., system-wide blackouts, are mitigated through the formation of stable islands. Compared to traditional approaches that base splitting decisions on local measurements at the locations of the relay [11], a WAMS with time-tagged PMU measurements allows a more effective splitting decision to be made based on multiple synchronized remote measurements used for state estimation. In the literature, different approaches to predict instability based on the system state are reported. Direct methods [12] evaluate the transient energy function based on the system state and compare it to a critical energy threshold to predict system stability. Machine-learning techniques are also used to assess transient stability. In [13] a decision tree (DT)-based tool is proposed to recognize conditions that trigger a grid-splitting action due to impending instability. In [14] , artificial neural networks (ANNs) are used to predict instabilities. The developed method is demonstrated on the IEEE 39-Bus Test System in combination with grid splitting and under frequency load shedding as mitigation actions. Other approaches are based on empirical relations, and activate splitting actions if the voltage phase angle differences and the derivatives of their signals exceed certain thresholds [15] . In order to find the appropriate splitting locations, several algorithms are proposed in the literature. A two-phase ordered binary decision diagram (OBDD) method is presented in [16] , which is based on a simplified graph and finds islands with coherent generators and a low power imbalance reducing the amount of load that needs to be shed after grid splitting. The two-step spectral clustering controlled islanding (SCCI) algorithm [17] finds groups of coherent generators and island solutions with a minimal power flow disruption minimizing the change of the power flow pattern following the grid splitting. The successful application of grid splitting depends on the communication infrastructure to collect system-wide synchronized measurements for state estimation based on which system stability is assessed. Furthermore, the communication infrastructure is necessary to relay the command to open line switches. However, communication can be degraded in terms of signal accuracy and delay. Signal accuracy is particularly important for measurement signals used for state estimation in order to successfully detect instability. Time delays experienced during the transmission of signals, however, are significant for both measurement and line switch opening signals because they affect the overall delay in grid splitting after the occurrence of the fault. The objective of this paper is to investigate the effects stemming from degraded communication on the successful application of grid splitting. A multi-machine power system including models of the generators, the electric network and the loads is used to simulate operating conditions, under which a loss of synchronism between generators following a component fault occurs. A model of the communication time delay for the transmission of measurements and line switch opening signals is applied with different communication network parameters to simulate the necessary flow Electronics 2016, 5, 49 3 of 25
doi:10.3390/electronics5030049 fatcat:dmvsa6fikzcjncfugp4htgupdq