CRII: CPS: Architecture and Distributed Computation in the Networked Control Paradigm: An Autonomous Grid Example
pdf
Abstract:
This project is focused on developing a fundamental understanding of the impact of network delays and data drops using an approach that is applicable to a variety of Cyber-Physical Systems (CPS). An example of such a CPS is the power grid which includes large-scale deployment of distributed and networked Phasor Measurement Units (PMUs) and wind energy resources. In this work, a 68-bus dynamic equivalent of New England – New York power system with 15 synchronous generators and 1 Doubly-Fed Induction Generator (DFIG)-based Wind Farm (WF) has been considered. Each generator and the WF are represented by a 6th order subtransient model and a 15th order averaged model, respectively; whereas the AC network is modeled algebraically. To perform modeling adequacy studies, two types of WF models are considered: a) Type I model that includes the Phase Lock Loop (PLL) and the Grid-side Converter (GSC) dynamics of WF, and b) Type II model, which neglects those dynamics. Gilbert-Elliot model is used to represent packet drop in the communication network. Modeling adequacy is performed through nonlinear simulation studies on this power grid with 151 dynamic states for stabilizing the electromechanical oscillations using remote PMU signals under different data dropouts rates, fault locations, and different degrees of nonlinearity. A centralized control architecture is assumed where the output of the controller is also sent over a communication network to the remote WF. Initial results from the modeling adequacy studies reveal that the Type II model is adequate in representing the system dynamics under a lower degree of nonlinearity, irrespective of the data-drop rates. However, with a high degree of nonlinearity the adequacy becomes dependent on the data drop rates. In absence of any data loss, Type II model is inadequate in representing the dynamics of the system. Interestingly, it is observed that this model produces more accurate response when high data dropout occurs in the communication network. Our ongoing research is aimed at an extensive evaluation of these results with further modeling adequacy studies. This will enable transformative Wide-Area Measurement Systems research for the smart grid by indicating the accuracy and the limitations of models stemming from the interaction among non-idealities in the communication network and the nonlinearity in the physical systems.
Submitted by Nilanjan Ray C…
on
Abstract:
This project is focused on developing a fundamental understanding of the impact of network delays and data drops using an approach that is applicable to a variety of Cyber-Physical Systems (CPS). An example of such a CPS is the power grid which includes large-scale deployment of distributed and networked Phasor Measurement Units (PMUs) and wind energy resources. In this work, a 68-bus dynamic equivalent of New England – New York power system with 15 synchronous generators and 1 Doubly-Fed Induction Generator (DFIG)-based Wind Farm (WF) has been considered. Each generator and the WF are represented by a 6th order subtransient model and a 15th order averaged model, respectively; whereas the AC network is modeled algebraically. To perform modeling adequacy studies, two types of WF models are considered: a) Type I model that includes the Phase Lock Loop (PLL) and the Grid-side Converter (GSC) dynamics of WF, and b) Type II model, which neglects those dynamics. Gilbert-Elliot model is used to represent packet drop in the communication network. Modeling adequacy is performed through nonlinear simulation studies on this power grid with 151 dynamic states for stabilizing the electromechanical oscillations using remote PMU signals under different data dropouts rates, fault locations, and different degrees of nonlinearity. A centralized control architecture is assumed where the output of the controller is also sent over a communication network to the remote WF. Initial results from the modeling adequacy studies reveal that the Type II model is adequate in representing the system dynamics under a lower degree of nonlinearity, irrespective of the data-drop rates. However, with a high degree of nonlinearity the adequacy becomes dependent on the data drop rates. In absence of any data loss, Type II model is inadequate in representing the dynamics of the system. Interestingly, it is observed that this model produces more accurate response when high data dropout occurs in the communication network. Our ongoing research is aimed at an extensive evaluation of these results with further modeling adequacy studies. This will enable transformative Wide-Area Measurement Systems research for the smart grid by indicating the accuracy and the limitations of models stemming from the interaction among non-idealities in the communication network and the nonlinearity in the physical systems.