University of Thessaly (UTH)
Energy management and consumer modeling in smart grid systems
Abstract
dc:descriptionIn the last decades, energy efficiency has turned into a major research issue, since the energy requirements of modern societies are growing continuously. Researchers have focused on optimizing the efficiency of the modernized power grid, i.e., the smart grid, which has evolved into a complex ecosystem with different actors such as consumers, operators and generators having different active roles in the system. They have also focused on improving the energy efficiency of ICT which is a prerequisite for efficient smart grids. Moreover, the advance of key technologies such as energy storage, renewable energy sources (RESs), communication and control has opened the way to new research directions. In this thesis, we present some key research problems in the context of this area, and we explore the use of control and optimization methods toward approaching them. Specifically, our goal is to address i) system challenges pertaining to the integration of energy storage and RESs into the smart grid and ii) challenges related to the energy consumer aspect. Energy storage devices, like uninterrupted power supply (UPS) or batteries, and Plug-in Hybrid Electric Vehicles (PHEVs), are prime resources for smart-grid efficiency improvement which need to be appropriately managed and incorporated into smart grid systems. In this context, we study two fundamental problems in energy storage management and dimensioning in smart grids. Specifically, first, we introduce the optimal energy storage control problem faced by an energy supplier, which amounts to deciding when and how much to charge and discharge a storage device in order to achieve a certain optimization objective in terms of energy generation cost. We address the problem above, first for a single storage device, and then for multiple storage devices that are shared among multiple micro-grids. Our optimization objective leads us to policies which attempt to keep balanced grid power consumption at all times. Next, we study a joint energy storage placement, dimensioning and management problem, given an available storage budget,where the goal is to minimize the power generation cost. We are interested in the way storage capacity placement and control impact the overall cost of energy generation. The solution policy for this problem involves various parameters such as the demand profiles of consumers, and power flow and balance constraints.RESs rely mainly on energy that flows naturally through the environment on a continual but time-varying basis and have received major attention in the last years due to their significant potential in reducing the carbon footprint. In this thesis, we study the use of renewable energy sources and scheduling techniques toward reducing the carbon footprint induced by an ICT system, namely, the web crawling component of a web search engine. This component discovers and downloads new pages on the Web as well as refreshes previously downloaded pages in the web repository. We introduce the problem of green web crawling, where the objective is to devise a page refresh policy that keeps the web pages as much as possible fresh and the carbon emissions that the web crawling process incurs on remote web servers low enough. We devise an optimal policy, which can be implemented in an online fashion, based only on the type of energy consumed by the servers and the staleness of the pages in the web repository. We also devise heuristics along the lines of the optimal policy and study their performance through experiments with real data. Demand-side management (DSM) activities for the smart grid aim to reduce or smooth enenergy consumption by providing dynamic prices or incentives in the form of monetary or non-monetary rewards to consumers. Different DSM techniques have been proposed, however, a fundamental issue in their design and operation is the recruitment of users. Serious-games design is an emerging area that can address precisely the issue of maximizing user engagement in various contexts. In this thesis, we discuss the use of serious games for demand-side management in smart grids. We introduce the problem of optimal serious-games design for the purpose of enforcing prudent energy consumption. We present a mathematical model of a simple gamification mechanism through which a serious-game designer (e.g., a demand-side management entity) aims to motivate consumers to reduce their energy consumption at peak hours by setting up a contest and by providing them incentives in the context of a serious game. The game designer optimally selects the game parameters, so as the utility-maximizing choices of consumers to minimize the energy generation cost of the energy supplier. We demonstrate that even such a simple serious-game design can provide adequate incentives to users for engaging in demand-side management. Finally, most proposed DSM schemes assume that consumers are rational decision makers which select their actions by solving complex optimization problems. However, consumers are humans and their decisions, which are driven by different factors, are far from rational. In this direction, we discuss the role of data in building behavior-based models for profiling energy consumers and predicting their behavior in DSM programs and energy consumption curtailment campaigns. Our ultimate goal is to discover through these models the different factors that determine consumer actions and the different importance placed on them. We present two different data-driven approaches for consumer modeling which are based on a popular machine-learning tool and a cognitive heuristic. We show that both approaches succeed in capturing the different importance that each consumer places on different factors and the uncertainty on consumer actions. We also introduce the optimal load-reduction task and incentive allocation problem faced by the designer of an energy consumption reduction campaign. The aim of the designer is to target tasks and incentives appropriately based on the different consumer profiles, so as to best fulfill the purpose of the campaign.
Degree
thesis:*- Grantor dc:publisher
- University of Thessaly (UTH)
- Year dc:date
- 2017
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Hatzi, Vassiliki
- Χατζή, Βασιλική
Subjects
dc:subject × 6Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/39977