Publication Date

12-2020

Date of Final Oral Examination (Defense)

12-11-2020

Type of Culminating Activity

Dissertation

Degree Title

Doctor of Philosophy in Computing

Department

Computer Science

Supervisory Committee Chair

Hoda Mehrpouyan, Ph.D.

Supervisory Committee Co-Chair

John Gardner, Ph.D.

Supervisory Committee Member

Amit Jain, Ph.D.

Supervisory Committee Member

Stephen Reese, P.E.

Abstract

Modern society has become increasingly reliant on the functioning of critical infrastructure. It is considered so vital that its incapacitation or destruction would have debilitating effects on the global economy, national security, and public health and safety. The electrical power system is uniquely positioned, as it is essential for all other sectors of critical infrastructure to operate as intended. However, it is constantly at risk due to factors such as natural disasters, climate change, aging infrastructure, and cyber threats. Thus, ensuring the efficient and continuous supply of electricity is of utmost importance and the topic of this dissertation.

The work in this dissertation covers two main topics; first the identification of a potential cyber threat to control system, and second, the foundation for a resilience framework to ensure a continuous supply of electricity in the grid.

Technology advancements have resulted in the integration of digital instrumentation and computational control through communication networks. This has resulted in systems which are more responsive, precise, reliable, and efficient. However, they are integrated into operational technologies without the necessary security defense. Designing an effective, layered security defense is not possible unless security threats are identified through a structural analysis of the control system. For that reason, an attacker's point of view is given for the reconnaissance effort necessary to gather details of the system dynamic that are required for the development of sophisticated attacks. A reconnaissance approach is presented that uses the system's input and output data to infer the dynamic model of the system. In this effort, a novel cyber-attack that targets the controller proportional-integral-derivative gain values in a constant setpoint control system is proposed. These findings will help researchers design more secure control systems.

The electrical power grid has been designed to withstand single component failures based on a set of reliability metrics that have proven acceptable during normal operating conditions. However, in recent years there has been an increasing frequency of extreme weather events. Many have resulted in widespread long-term power outages, proving reliability metrics do not provide the adequate energy security that is needed.

As a result, researchers have focused their efforts on resilience metrics to ensure efficient operation of power systems during extreme events. A resilient system has the ability to resist, adapt, and recover from disruptions. Therefore, resilience is a promising concept for the current challenges facing power distribution systems.

An operational resilience metric for modern power distribution systems is presented. The metric is based on the aggregation of system assets adaptive capacity in real and reactive power. The metric indicates the control limits of the assets of the system. This also relates to the magnitude and duration of a disturbance the system can withstand. The mathematical details of the metric are covered and consider the real-time operational outputs of the assets, its ramp rates, latency, and energy limits. The metric is then focused on the resilience contribution of the three types of hydropower generation and their contribution to the various time scales or ``Rs" of resilience. Further analysis demonstrates using very short-term (seconds) and short-term (day-long) solar PV generation forecast with uncertainty. It was demonstrated that the addition of battery storage to a solar generation asset can be used to maintain adaptive capacity during times where solar generation is at the negative uncertainty scenario.

DOI

10.18122/td/1768/boisestate

Download

Share

COinS