Abstract: Cascading power outages can inflict devastating consequences on the economic and operational security of city infrastructure, and can even result in fatalities. Service loss of a single transmission line can cause the subsequent shutdown of topologically neighboring, or electrically dependent, transmission lines as power is rerouted to meet customer demand. When apparent power demand exceeds the carrying capacity of a neighboring transmission line, that line will shut down, and a chain-reaction of propagating transmission line outages will commence, leading to a widespread outage. When a transmission line fails, the downstream network of distribution lines that deliver power to individual homes and businesses will shut down, causing loss of power to critical appliances, such as refrigeration systems, which require constant power to prevent food spoilage. The Great Blackout of 2011 on September 8 affected seven million California residents, and spread into Arizona and parts of Baja California in Mexico. Caused by the unintentional mistake of a single technician who shut down a 500 kV transmission line between the Hassayampa and North Gila substations in Arizona, the outage propagated through five major power grids within 11 minutes, which resulted from the sequential overload of networked transmission lines. The outage cost homeowners and restaurants between $12 and $18 million in damages from the spoilage of perishable food. During the Northeast blackout of 2003, nearly 100 people died, with one death resulting from loss of air conditioning that was required to keep the skin grafts of a burn victim adequately cooled, thereby illustrating the need for Smart Cities to maintain power to critical appliances as load is shed during propagating outages. The Northeast blackout affected 45 million people in the US, 10 million people in Ontario, and was caused by a software error that failed to re-distribute load to other high-capacity transmission lines after a set of 345 kV transmission lines in Walton Hills Ohio ground faulted by sagging into underlying trees. US electricity consumers lose an incredible $79 billion dollars each year from combined power outages, with a single large outage costing in the order of $10 billion dollars. In this talk, we present an LPWAN fog-computing framework called MIST, designed to monitor and measure local consumer power demand and perform real-time load- shedding decisions at the network edge. Should a localized failure occur, the fog network reacts to mitigate failure propagation by shedding load in a structured way to reduce consumer loss by powering off load in priority order. The MIST framework consists of "smart" electrical receptacles with micro controller measurement devices to monitor and predict the transient coverage of a transmission line outage and cooperatively sheds power to halt propagation. MIST receptacles use the LoRa chirp spread spectrum radio modulation technology for LPWAN, which uses the license-free sub-gigahertz (915 MHz in US) band. Current power shedding techniques employed by energy companies do not distinguish among user home appliances, and can result in a total power loss to selected users. The proposed MIST fog- computing architecture will benefit Smart Cities by providing a means by which energy companies can selectively shed power at the very edge of a grid by powering off customer appliances in order of increasing criticality, thereby mitigating the propagation of a failure and reducing consumer costs that result from power loss.
Bio: Christopher Paolini is an Assistant Professor in the Department of Electrical and Computer Engineering at San Diego State University. Chris is the recipient of grants from the Department of Energy and NASA, and five NSF Office of CyberInfrastructure awards, most recently the current NSF CC* Grant 1659169 “CC* Storage: Implementation of a Distributed, Shareable, and Parallel Storage Resource at San Diego State University to Facilitate High-Performance Computing for Climate Science”. Christopher Paolini's current research interests include Internet of Things device development, machine learning, embedded systems, cloud computing, big data analytics, deep learning, software engineering, numerical chemical thermodynamics, numerical chemical kinetics, numerical geochemistry, high performance computing, scientific computing and numerical modeling, high speed (100gbps) networking, cyber infrastructure development, and cybersecurity. Chris received a B.S. degree in Computer Science in 1991, M.S. degree in Computer Science in 1998, and his Ph.D. degree in Computational Science in 2007, all from San Diego State University.