The Two-Edged Sword of Democratizing Grid Control: Distributed Inertia Control as a Cyber-Physical Attack Path

Among emerging trends in power systems today is the "democratization" of control; that is, the potential for vast numbers of distributed, consumer-based devices to participate in control to improve grid dynamic performance. Coincident is an associated trend towards power-electronically-coupled renewable resources, displacing traditional synchronous generators, and reducing the desirable effects of rotating inertia on the grid. A natural fix to the later problem presents itself in the merger of these two trends: use of "emulated inertia" control, implemented in distributed, consumer-based loads. The work here is intended as a cautionary note, to emphasize the strong need for cyber-security in any such control implementation. Emulated inertia feedback is extremely well suited as a path of cyber-attack. We will demonstrate that such control on loads can be subverted into destabilizing "bots," creating wide-area instabilities with only slight modification of feedback parameters, and no real-time communication between local controllers. The amount of affected load can be relatively modest, and the attack can be designed to selectively target specific generators to experience large electromechanical oscillations. These oscillations would be of sufficient magnitude to trip rate-of-change-of-frequency protective relays, and could trigger cascading disconnection of generators. This scenario may be of particular concern, because today's grid cyber-security standards appear ill suited to consumer level devices (e.g., plug-electric vehicle chargers).
My research interests center on nonlinear stability and control theory, with particular emphasis on applications in electric power systems.My work in this area has focused on a number of topics. The first was the problem of identifying region-of-attraction estimates in systems described by ordinary differential equations with algebraic constraints, sometimes referred to as "semi-implicit" systems, but only recently have received attention in the literature on direct methods of transient stability analysis.Region-of-attraction estimates have important practical application in determining the ability of a power system to recover from line faults, load disturbances , and loss of generation. A second area of research in electric power systems involves illuminating the relationship between stability properties of deterministic models for the power system and exit-time statistics in a stochastic differential equation model having broad-band random disturbances in the load term. Using results from optimal control theory, we have demonstrated a closed form expression for the so-called "quasi potential" function for the power system, and shown its relationship to a standard deterministic Lyapunov function.This result offers considerable promise in yielding practically implementable algorithms to indicate in real time the vulnerability of the power system to such phenomena as voltage collapse. A third area of my research that ties developments in modern control theory with power systems applications, is the topic of "robust stability."a power system is a highly nonlinear system, whose operating point is continually changing in time.We are extending results in robust control theory to obtain a means of guaranteeing stable operation of the system over a whole range of operating points, and examining the development of algorithms for application to large scale systems.