Distributed integrated energy management systems in residential buildings

Highlights

• A home energy management system with multiple decision-making units (energy managers) is proposed.

• Each energy manager represents a household device.

• The decision making procedure is based on a discovery-and-negotiation-mechanism.

• Load shifting schedules within user specified periods is the outcome of the system.

• The system is able to save operational energy costs of a household.

Abstract

This paper explains the concept of a distributed integrated energy management (diEM) system for residential buildings. The overall goal of the system is to minimize operational energy costs of the household. This is obtained by load shifting in order to enhance the self-consumption rate of the on-site renewable electricity production. The crucial difference to centralized energy management systems (where data from household devices must be gathered and evaluated centrally) is the presence of multiple smart energy managers that negotiate with each other on the switch-on times of their dedicated electric devices. The major benefit is that devices of various manufacturers can be incorporated in the same home energy management system with open standards and open protocols without any additional decision-making unit. The basic procedure of the diEM is divided into a discovering phase and a negotiation phase. Best practice parameter settings are deduced from realistic scenarios with constant and variable electricity tariffs, and a run-time analysis indicates that up to seven devices can run simultaneously with a one minute renegotiation frequency. The monetary evaluation shows that the diEM can reduce the operational energy costs at a rate dependent on user behavior and tariff structures.

Introduction

Due to its nature, solar driven photovoltaic (PV) modules feed heavily fluctuating power into the grid over time. The ongoing rollout of renewable capacities worldwide [1] is leading to an increase in situations in which the grid or parts of it are oversupplied or undersupplied. Conventional power plants or energy storage systems must balance the grid in these cases. Another way to deal with this problem is demand-side management (DSM) at the electricity consumer, for example in residential buildings [2]. One measure to employ DSM and to cut feed-in peaks of renewables is to incentivize self-consumption of on-site produced electricity in microgrids that are connected to the low voltage grid [3]. The incentive mechanism is feed-in tariffs which are lower than purchase tariffs. The German Renewable Energy Sources Act (2014) may serve as an example for that mechanism, since the feed-in tariff for rooftop systems of less than 10 kWp (launched in February 2015) accounts for 12.92 €ct/kWh [4] and the electricity purchase tariff averaged out at 28.81 €ct/kWh [5]. If home and PV-system owners want to act in an economical way, they can install a home energy management system (HEMS) that takes over the decision-making process, when to switch on or off, for many household devices. Those devices with the highest power and energy consumption should be operated by the HEMS. Most such systems employ a central unit with varying levels of computational power [2], [6], [7], [8]. This unit collects all necessary data from connected household devices and calculates an optimal strategy for the overall load curve. Different methods have been developed in recent years to generate an optimized schedule. Some researchers use binary linear optimization [8], mixed-integer linear (MIL) or nonlinear optimization [9], [10], [11], while others apply genetic algorithms [12] or heuristics [11], [13]. Multi-objective optimization is also an option in order to reduce the number of optimization variables [14]. However, all systems have in common that they are centrally organized and operated, which comes with some drawbacks. For example, high complexity with many decision variables in MIL programming requires more computational time, setting limits for recalculating schedules when any input parameter has changed [11].

In contrast, this paper discusses a decentralized approach: A distributed integrated energy management system (diEM), where all household devices come with a computational unit. These components are called energy managers, which detect each other in the network [15] and negotiate with each other on available PV-power in order to cut household energy costs. The idea is to have manufacturer independent devices connected in an open environment with open interfaces and open protocols. This should lower development costs for enabling devices to cooperate in a HEMS. It is up to each manager to acquire any external information necessary to produce optimized results. Household appliance manufacturers do not have to share their optimization know-how with a central decision-making unit. Instead, only a limited amount of information, like load or price profiles, must be exchanged between managers. In comparison to a centralized system, the diEM is more robust against manager failure. If one manager drops out of the process, the remaining managers can still do their jobs independently and achieve an optimized solution within the remaining community. If the decision making unit in a centralized system breaks down, however, the total energy management system fails.