Program Overview
08:45 Opening Remark (Mariagrazia Dotoli, Sergio Grammatico)
08:50 Sergio Grammatico “Population Control for Large-Scale Smart Energy Systems”
09:25 Emanuele Crisostomi “From Cyber-Physical Systems to Internet of Things: Smart Procurement of Naturally Generated Energy”
10:00 Peter Luh “A Synergistic Combination of Surrogate Lagrangian Relaxation and Branch-and-Cut for Mixed-Integer Optimization Problems”
10:35 Coffee Break
10:55 Raffaele Carli “Decentralized and Distributed Optimization For Energy Scheduling of Interconnected Smart Homes”
11:30 Kostas Margellos “Decentralized and Distributed Building Control in the presence of Uncertainty”
12:05 Closing Remark (Mariagrazia Dotoli, Sergio Grammatico)
In this workshop we bring together academic experts in the areas of control and optimization for energy management of large-scale smart systems. We will focus on scalable decentralized and distributed methods for the optimal energy management of systems employing intelligent controllers. The workshop will start from a game-theoretical population control approach for the coordination of locally optimal decisions, with specifi c application to the problem of charging large fleets of plug-in electric vehicles and the demand response management of smart building aggregations. The plug-in electric vehicle application will be further addressed via distributed algorithms, accounting for the transportation mobility patterns, the local energy management units and the power grid, to prioritise the employment of renewables, in the presence of uncertainty sources. Decentralized and distributed (stochastic) optimization will be then discussed as methodologies to coordinate the energy management systems in aggregations of interconnected smart homes/buildings, in the presence of shared renewable sources and uncertain weather conditions. The workshop will also present some recent advances in the fi eld of mixed-integer optimization for large-scale energy systems, with application to the unit commitment problem for power systems. The workshop will conclude with a round-table discussion about the future of research in the broad area of optimal control and coordination for large-scale smart energy systems.
Topics and descriptions
Population Control for Large-Scale Smart Energy Systems
Dr. Sergio Grammatico.
Large-scale energy systems consist of many heterogeneous agents, interacting econom ically and physically, including power generation plants, distributed local generation and storages, large and small consumers, e.g., wind farms, plug-in electric vehicles and energy ecient buildings, respectively. These energy systems are usually managed independently, according to di erent economic and operating objectives. However, the dynamic interaction of the locally managed agents gives rise to complex emerging behaviors that can lead to undesired market price fluctuations and large-scale disruptions in the electric grid. The talk presents game-theoretic methodologies for large-scale smart energy systems, where individual, as well as aggregations of, agents employ intelligent controllers to optimally plan strategic decisions, possibly based on forecast information about the future economic and physical environment. Applications to energy demand management for aggregations of plug-in electric vehicles and smart buildings are shown.
From Cyber-Physical Systems to Internet of Things: Smart Procurement of Naturally Generated Energy
Prof. Emanuele Crisostomi.
Electric vehicles are an attractive means of transportation for improving air quality, especially if they are powered by electricity generated from natural gas or wind, water or solar power. In this talk we illustrate an innovative vision that takes into simultaneous account transportation mobility patterns, the energy management units of Plug-in Hybrid Electric Vehicles (PHEVs) and the power grid, to prioritise energy generated from renewable source to recharge the batteries of the vehicles. In this way, ICT technologies are exploited to prevent green energy from being wasted and at the same time to mitigate the uncertainty of the power grid to accommodate unexpected electrical loads.
A Synergistic Combination of Surrogate Lagrangian Relaxation and Branch-and-Cut for Mixed-Integer Optimization Problems
Prof. Peter B. Luh.
Mixed-integer optimization problems are prevalent in dynamic systems. Historically, Lagrangian relaxation and subgradient methods were used to solve such problems by exploiting separability through relaxing coupling constraints and decomposing relaxed problems into subproblems. Although subgradient methods have been widely used to update multipliers, they require the relaxed problem to be fully optimized, and this can be computationally expensive. Moreover, convergence can be slow because multipliers often zigzag across “ridges” of the dual function. The recent trend is to solve these problems by using the branch-and-cut method that exploits problem linearity. However, complex problems such as stochastic unit commitment or even the deterministic version with combined cycle units in power systems can pose major computational challenges. The reason is that cuts that defi ne a convex hull, referred to as facet-de fining or strong cuts, are problem-dependent and may be dicult to obtain. Complex transitions among combined cycle states associated with one unit, for example, are treated as global constraints, and a ffect the entire problem. Very recently, we developed the Surrogate Lagrangian Relaxation method where a proper direction to update multipliers can be obtained without optimally solving all subproblems with much reduced computational eff ort and multiplier zigzagging. More importantly, convergence to the optimum does not require the knowledge of the optimal dual value. This was achieved with a constructive process in which distances between Lagrange multipliers at consecutive iterations decrease, and as a result, multipliers converge to a unique limit. The decrease cannot be too large to avoid premature termination of the iterative updating process. To enable an ecient exploitation of separability as well as linearity, surrogate Lagrangian relaxation and branch-and-cut are synergistically combined where surrogate Lagrangian relaxation is used to decompose a problem into subproblems by relaxing coupling constraints, and each subproblem is solved by using branch-and-cut. With decomposition, the complexity of a subproblem is much smaller than that of the original problem. Constraints associated with a subproblem are handled locally and no longer a ffect the entire problem. Furthermore, by exploiting the novel observation that subproblem constraints and therefore the associated subproblem convex hulls remain unchanged after multipliers are updated, solving subproblems becomes much easier than starting from scratch. Numerical results on unit commitment with combined cycles demonstrated that the new approach is computationally efficient. The approach opens up a new direction for optimizing mixed integer optimization problems in power systems and beyond.
Decentralized and Distributed Optimization For Energy Scheduling of Interconnected Smart Homes
Mr. Ra aele Carli.
Concerns on energy consumption, both in terms of sustainability and resource exhaustion, are leading towards efficient energy behaviors all smart cities actors, including small end-users such as homes. From its initial focus on residential comfort, the smart home concept is now focusing on the development and installation of new technologies and of e ffective and efficient energy management systems. Hence, smart homes are proactive customers (prosumers) that negotiate and collaborate as an intelligent network in close interaction with their external environment. In this context, the talk deals with the optimal scheduling of energy activities of a group of interconnected users equipped with controllable electrical appliances, renewable energy sources, dispatchable energy generators, and energy storage systems. Having access to pricing information, residential energy controllers at fi rst provide the householder with the ability to minimize the cost of energy by suitably scheduling energy activities and shifting energy purchase from the peak time or high price periods to low use or reduced cost periods. As a consequence of increasing shares of renewable energy and of the improving performance of small scale storage systems, the smart homes energy scheduling problem is being focused on solving considering the e ective integration of distributed generation and storage. Thus, the objectives of optimization of residential energy activities are recently extended to achieving the maximum exploitation of energy locally produced by renewable sources and the most convenient management of storage charging/discharging strategies. The talk
presents decentralized and distributed optimization techniques, such as game-theoretic methodologies, for large-scale residential energy systems, where individual, as well as aggregations of, smart users make use of energy scheduling systems to optimally manage the use of electrical appliances, plan the energy production and supplying, and program the storage systems charging/discharging. Applications are shown to demonstrate that the presented approached allow full exploitation of the potential of local energy generation and storage to reduce the individual user energy consumption costs, while limiting the peak average ratio of the energy pro les and complying with the customers’ energy needs.
Decentralized and Distributed Building Control in the presence of Uncertainty
Dr. Kostas Margellos.
Abstract: Building control has attracted a considerable research attention the last years due to the signi cant energy savings of an intelligent building management scheme. In most cases, however, weather conditions and building occupancy are not accurately known, rendering building control a stochastic energy management problem. Stochastic optimization techniques o er a powerful tool to tackle such problems, but adopting a centralized control paradigm imposes communication and computation challenges. In this talk we address these issues and discuss decentralized and distributed control architectures to solve the building control problem. Both approaches rely on an iterative methodology, where every building solves a stochastic but local energy management problem, followed by a communication round of information exchange with other buildings. In the distributed control approach, information has to be exchanged only with neighboring buildings, thus leading not only to computation but also communication savings. We present a novel iterative
algorithm for distributed, stochastic optimization, analyze its convergence and optimality properties, and evaluate its ecacy on a building control set-up.
Organizers: Dr. Sergio Grammatico, ETH Zurich, Switzerland. and Prof. Mariagrazia Dotoli, Polytechnic
University of Bari, Italy.
Speakers
Dr. Sergio Grammatico, ETH Zurich, Switzerland
Sergio Grammatico was born in Marsala, Italy, in 1987. He received the B.Sc. in Computer Engineering in 2008, M.Sc. and Ph.D. degrees in Automation Engineering in 2009 and 2013, respectively, all from the University of Pisa, Italy. He also received a M.Sc. degree in Engineering Science from the Sant’Anna School of Advanced Studies, Pisa, Italy, in 2011. He visited the Department of Mathematics at the University of Hawai’i at Manoa in 2010 and 2011, and the Department of Electrical and Computer Engineering at U.C. Santa Barbara in 2012. Since 2013 he is a post-doctoral Research Fellow at the Automatic Control Laboratory, ETH Zurich, Switzerland. He was nominated IEEE TAC Outstanding Reviewer in 2013 and 2014. His research interests include Lyapunov control systems, stochastic, game-theoretic control and optimization, with application to large-scale smart energy systems.
Prof. Emanuele Crisostomi, University of Pisa, Italy
Emanuele Crisostomi received the B.S. degree in computer science engineering, the M.S. degree in automatic control, and the Ph.D. degree in automatics, robotics, and bioengineering, from the University of Pisa, Italy, in 2002, 2005, and 2009, respectively. He is currently an Assistant Professor of electrical engineering with the Department of Energy, Systems, Territory and Constructions Engineering, University of Pisa. His research interests include control and optimization of large-scale systems, with applications to smart grids and green mobility networks.
Prof. Emanuele Crisostomi, University of Pisa, Italy
Emanuele Crisostomi received the B.S. degree in computer science engineering, the M.S. degree in automatic control, and the Ph.D. degree in automatics, robotics, and bioengineering, from the University of Pisa, Italy, in 2002, 2005, and 2009, respectively. He is currently an Assistant Professor of electrical engineering with the Department of Energy, Systems, Territory and Constructions Engineering, University of Pisa. His research interests include control and optimization of large-
scale systems, with applications to smart grids and green mobility networks
Prof. Peter B. Luh, University of Connecticut, USA
Peter B. Luh received his B.S. from National Taiwan University, M.S. from M.I.T., and Ph.D. from Harvard University. He has been with the University of Connecticut since 1980, and currently is the SNET Professor of Communications & Information Technologies. He was the Head of the Department of Electrical and Computer Engineering from 2006 to 2009. He is also a member of the Chair Professors Group, Center for Intelligent and Networked Systems (CFINS) in the Department of Automation, Tsinghua University, Beijing, China. Professor Luh is a Fellow of IEEE. He was the VP of Publications of RAS (2008-2011), the founding Editor-in-Chief of the IEEE Transactions on Automation Science and Engineering (2003-2007), and the Editor-in-Chief of IEEE Transactions on Robotics and Automation (1999-2003). He received IEEE Robotics and Automation Society 2013 Pioneer Award for his pioneering contributions to the development of near-optimal and eefficient planning, scheduling, and coordination methodologies for manufacturing and power systems. His research interests include Smart Power Systems smart grid, design of auction methods for electricity markets, robust renewable (wind and solar) integration to the grid, and electricity load and price forecasting; Intelligent Manufacturing Systems planning, scheduling, and coordination of design, manufacturing, and service activities; Smart and Green Buildings and Eco Communities optimized energy management, HVAC fault detection and diagnosis, emergency crowd guidance, and ecocommunities.
Mr. Ra aele Carli, Polytechnic University of Bari, Italy
Ra aele Carli received the Laurea degree in Electronic Engineering with honours in 2002 from Politecnico di Bari, Italy. He is currently pursuing the Ph.D. degree in Electrical and Information Engineering at Politecnico di Bari, Department of Electrical and Information Engineering, Italy in the framework of “Res Novae” Project supported by the “Smart Cities Communities and Social Innovation” program of the University and Research Italian Ministry. From 2003 to 2004, he was a Reserve Ocer with Italian Navy. From 2004 to 2012, he worked as System and Control Engineer and Technical Manager in the Space and Defence sector. His research interest includes de nition and simulation of decision and control systems and modeling and optimization of complex systems. Dr. Kostas Margellos, Polytechnic University of Milan, Italy
Kostas Margellos (S’09M’14) received his diploma in Electrical and Computer Engineering from the University of Patras, Greece in 2008, and the Ph.D. degree in automatic control from the Department of Information Technology and Electrical Engineering, at ETH Zurich, Switzerland, in 2012. He spent 2013 and 2014 as a post-doctoral researcher in the Automatic Control Laboratory at ETH Zurich and in the Department of Industrial Engineering and Operations Research at UC Berkeley, respectively. As of February 2015 he continues his post-doctoral research at the Polytechnic University of Milan. His research interests include optimization and control of complex uncertain systems, with applications to generation and load side control for power networks.
Dr. Kostas Margellos, Polytechnic University of Milan, Italy
Kostas Margellos (S’09M’14) received his diploma in Electrical and Computer Engineering from the University of Patras, Greece in 2008, and the Ph.D. degree in automatic control from the Department of Information Technology and Electrical Engineering, at ETH Zurich, Switzerland, in 2012. He spent 2013 and 2014 as a post-doctoral researcher in the Automatic Control Laboratory at ETH Zurich and in the Department of Industrial Engineering and Operations Research at UC Berkeley, respectively. As of February 2015 he continues his post-doctoral research at the Polytechnic University of Milan. His research interests include optimization and control of complex uncertain systems, with applications to generation and load side control for power networks.