Quantitative reliability indices are important to utility companies, vendors, and regulators for planning, operation, maintenance, and regulatory purposes. System reliability evaluation methodologies have been mostly focusing on the current-carrying part which is called physical part in this report and the cyber part is assumed to be perfectly reliable in these evaluations. As a part of the efforts to make the grid smarter, the cyber part has been rapidly expanding. The term “ cyber” refers to the devices and activities residing in the secondary side of the power system, associated with the functionalities of measurement, control, monitoring, and protection. The overall power system is also referred to as a “ cyber-physical power system” , in which the

communication networks and power components are interdependent. The cyber-physical interdependencies exist extensively in power systems at all levels and associate with various aspects. The main focus of this work is to investigate this cyber-physical interdependence and develop methods for evaluation of reliability considering this mutual relationship.

State estimation, and health monitoring for general situational awareness at a power system control center, inevitably employs a mix of a priori information, based on system model and parameter data, and observations in the form of measurement data. In traditional power systems applications, when SCADA systems provided only low-bandwidth, unsynchronized measurement data to a control center, observations and a priori information typically contributed on nearly equal footing in applications such as state estimation and topology error identification.

Today, the availability of high-bandwidth, time-synchronized PMU data shifts this balance, creating a much larger role for observations, and reducing the need for full model information, thereby opening the door to much faster time-scale estimation. However, rather than replacing existing state estimation tools and the applications that depend on them, the techniques developed in this project seek to provide much higher bandwidth estimation, to “ fill in” situational awareness between slower, periodic state estimator updates. The particular algorithms developed focus on two applications: (A1) fast time-scale, dynamic detection and identification of network topology changes; and (A2) PMU measurement-based, (nearly) “ model-free” estimation of the power flow Jacobian.

State estimation, and health monitoring for general situational awareness at a power system control center, inevitably employs a mix of a priori information, based on system model and parameter data, and observations in the form of measurement data. In traditional power systems applications, when SCADA systems provided only low-bandwidth, unsynchronized measurement data to a control center, observations and a priori information typically contributed on nearly equal footing in applications such as state estimation and topology error identification.

Today, the availability of high-bandwidth, time-synchronized PMU data shifts this balance, creating a much larger role for observations, and reducing the need for full model information, thereby opening the door to much faster time-scale estimation. However, rather than replacing existing state estimation tools and the applications that depend on them, the techniques developed in this project seek to provide much higher bandwidth estimation, to “ fill in” situational awareness between slower, periodic state estimator updates. The particular algorithms developed focus on two applications: (A1) fast time-scale, dynamic detection and identification of network topology changes; and (A2) PMU measurement-based, (nearly) “ model-free” estimation of the power flow Jacobian.

Power system dynamic simulations and transient stability studies vitally depend on the accuracy

of the load models. However, it is widely recognized that the monitoring capability for loads is

insufficient while the accuracy of load models has not been fully investigated. Meanwhile, the

analysis on the adequacy or necessity of existing dynamic composite load models is left open the

literature. This project aims to tackle these challenging by analyzing the complexity trade-off of

the load models used for studying the increasingly critical fault-induced delayed-voltage recovery

(FIDVR) events. Real-time and non-intrusive estimation of the load models are also pursued in this

project, by leveraging high-resolution and high-bandwidth data even available within distribution

feeders. This project has successfully developed a software tool for grid operators to perform

measurement-based load modeling using the WECC composite load model (CMPLDW) structure.

Power system dynamic simulations and transient stability studies vitally depend on the accuracy

of the load models. However, it is widely recognized that the monitoring capability for loads is

insufficient while the accuracy of load models has not been fully investigated. Meanwhile, the

analysis on the adequacy or necessity of existing dynamic composite load models is left open the

literature. This project aims to tackle these challenging by analyzing the complexity trade-off of

the load models used for studying the increasingly critical fault-induced delayed-voltage recovery

(FIDVR) events. Real-time and non-intrusive estimation of the load models are also pursued in this

project, by leveraging high-resolution and high-bandwidth data even available within distribution

feeders. This project has successfully developed a software tool for grid operators to perform

measurement-based load modeling using the WECC composite load model (CMPLDW) structure.