A New Way to Look at Energy and Data Flows
Written by Janet Roveda, Susan Lysecky and Young-Jun Son
The smart grid requires seamless integration of software and hardware components and can be viewed as a cyber-physical energy system that integrates information and energy flows. Accordingly, it makes sense to do research on the development of applications to monitor and manage those flows. Two such applications borrow from a theory developed to support task scheduling over the Internet with a vast number of servers.
A key roadblock, both to achieving electricity conservation and integration of renewable generation, is the high complexity of the underlying smart grid design. Tremendously numerous components, standards and data are required to provide a unified solution. Thus, how to model smart homes and how to effectively manage large number of households are key issues in smart grid design.
Take just the electronic devices and equipment found in today's smart home or the smart home of the near future. To the extent solutions are available they offer a variety of sensors that can be built in or attached to a number of household appliances. Such systems enable users to monitor appliances through websites or mobile devices, receiving sensor data in real-time. But these sensors currently concentrate on household energy monitoring and are not very "smart" because they do not provide users with many suggestions on how to save energy, and do not generally incorporate feedback to dynamically respond to the changing demands of the environment.
The smart grid requires seamless integration of software and hardware components and can be viewed as a cyber-physical energy system that integrates information and energy flows. Accordingly, we have focused our research on development of models and frameworks to monitor and manage those flows. We have devised two mechanisms, an Energy-Timing Interface Model (ETIM) and the Energy Management and Optimization (EMO) framework.
ETIM is a modeling concept inspired by the interface theory proposed by T.A. Henzinger and E.A. Lee, who are professors of electrical engineering and computer science at the University of California, Berkeley. Their goal was to invent a new information model that could support task scheduling over the Internet with a vast number of servers. To that end, each task is modeled as an interface box, where each task has its own duration, resource requirements and available resources. So, each interface box has a duration time as well as required and provided resources, as its basic entries.
Since task scheduling in the smart grid resembles the scenario that Henzinger, Lee and their colleagues addressed for the Internet, we adapted their interface theory to manage data and electricity flows. Their interface box is utilized to produce a compact model consisting of a set of boundary conditions, with parameters specifying requested power, output/response power, and delay (or timing).
In contrast to the original interface box model for the Internet domain, the ETIM model focuses on components for energy management rather than just data communication; the EMO framework is also different from its counterpart in the interface box and optimization framework. ETIM focuses on how to schedule tasks so they are supported by enough energy, while EMO emphasizes how to schedule tasks so that they can be processed in a certain required sequence taking data communication requirements into account.
EMO is a SystemC-based framework—a C++ class library used for hardware architecture—that integrates individuals' energy usage patterns, energy generation patterns and a number of scheduling methodologies. In addition, this framework includes algorithms to determine various configurations of components at each household, and is applicable on a different platform, such as a microgrid.
A noteworthy aspect of the ETIM model is its abstraction of the detailed physical and electrical properties of each component. That is, the model neglects a lot of detailed information and only cares about boundary conditions and timing. This procedure reduces the computational complexity for a given microgrid. Combined with the EMO framework and fast estimations of energy flows, ETIM enables sensor data reduction, keeping only information for abnormal cases, and fast estimations of energy flows. For instance, the model records and approximates energy provided by resources and energy required by components via a set of energy computation equations at every moment.
ETIM and EMO can perform a variety of tasks. For example, given a single household, the new ETIM checks whether its demand exceeds the supply at a given moment. If demand is greater than supply, grid energy will be called for to balance the household's needs.
Equally important, the applications will facilitate determination of how underlying hardware should best be configured. Instead of querying each appliance, the automated decision maker checks whether the interface boxes are compatible. That is, it finds out whether ETIM A can be connected with ETIM B by checking if the boundary conditions are satisfied.
Consider an exemplary household configuration where every component within the household is modeled by an interface box. If the sun is not providing sufficient energy at a given time, the interface box model for PV cells fails to produce the supply energy. Thus, there is no need to turn on converters. Instead, the decision maker will rely on batteries and/or the electrical grid to supply appliances.
This kind of compatibility check leads to data-driven dynamic reconfigurations of the household. At the grid level, another layer of management is needed for the smart grid. Consider a set of tasks requested by customers from multiple households. The grid level or community level EMO framework is needed to determine how to satisfy all users' requests using minimum grid energy. The required grid power/energy is provided from the decision maker from each household, along with a set of task requests. The output of the grid level EMO framework is a set of tasks with timing information and the priority.
The cost of ETIM and EMO hardware and installation must be minimal to ensure accessibility to a variety of end users. It should be limited to less than $200, which could be divided into small increments and incorporated into monthly bills. Moreover, the tools and features provided must encompass a number of customer needs available through various mediums such as computers, Internet servers, and/or cloud to provide real time monitoring and management. In addition to providing monitoring solutions, ETIM and EMO also have the potential to report in real-time abnormal activities in hardware components (PV cells, batteries, and appliances) and unusual usage patterns and can be addressed quickly. Moreover, privacy and security mechanisms must be considered.
Janet Roveda, a professor of electrical and computer engineering at the University of Arizona and IEEE senior member, earned a bachelor's degree in computer science at the East China Institute in Nanjing, China in 1991, and master's and doctoral degrees in electrical engineering and computer science from the University of California, Berkeley in 1998 and 2000. She was a recipient of the NSF career award in 2005, the PEACASE award in 2006, the University of Arizona Outstanding Achievement Award in 2007 and the R. Newton Graduate Research Project Award from the Design Automation Conference. Her primary research interests focus on smart grid circuit design, VLSI circuit modeling and analysis, and low power multi-core system design.
Young-Jun Son is a professor of systems and industrial engineering, a da Vinci Fellow, an Arizona Engineering Faculty Fellow, and Director of the Advanced Integration of Manufacturing Systems and Technologies Program at the University of Arizona. His research focuses on modeling and control of complex manufacturing and service enterprises, distributed federation of multi-paradigm simulations, and modeling human decision-making and social behaviors. A member of IEEE, he received the SME 2004 Outstanding Young Manufacturing Engineer Award, the IIE 2005 Outstanding Young Industrial Engineer Award, the IERC Best Paper Awards (in 2005, 2008 and 2009), and Best Paper of the Year Award in 2007 from IJIE.
Susan Lysecky is an assistant professor in the Department of Electrical and Computer Engineering at the University of Arizona, where she coordinates research efforts for the Ubiquitous and Embedded Computing Laboratory. She received her M.S. and Ph.D. degrees in computer science from the University of California, Riverside in 2003 and 2006, and is a member of IEEE and ACM. Her current research interests include embedded system design, with emphasis on self-configuring architectures, human-computer interaction and facilitating the design and use of complex sensor-based system by non-engineers.