Model-based dynamic relaying for power system protection under uncertainty

Several major cascading outages have involved mis-operation or mis-coordination of protective relays during stressed system conditions that resulted in a vulnerable network. Such stressed conditions include concurrent high load demand, changes in circuit topology, equipment outages, and short-circuit faults. With levels of wind and photovoltaic (PV) generation projected to increase in the future, large-scale variable generation also presents an additional point of vulnerability to the existing protection system. In this work, a new framework is introduced that is built on model-based distributed relay intelligence. The framework integrates real-time measurements from adjacent buses and predictive circuit models embedded in relays. The data collected by the relay is input to circuit simulations in order to accurately predict possible fault conditions at the relay location. Settings can then be adapted in real-time based on prevailing system conditions. Several scenarios are evaluated to demonstrate the effectiveness of this approach. This work further develops a probabilistic formulation of optimal relay characteristics that adapts to the randomness and uncertainty introduced by renewable generation. In this framework, the calculation of relay operating times is formulated as a stochastic optimization problem. In addition, at the system level, a mixed-integer linear program is developed for protective device and switch allocation considering intentional islanding with distributed generation in distribution systems.