Cooperative System and Scheduling Algorithm for Sustainable Energy-Efficient Communities

Abstract

The High connectivity among devices within the Internet-of-Things facilitates two-way flow of information throughout the infrastructure reaching homes and the consumers targeting broader energy goals. Our proposal encompasses consumers cooperating in response to utility supply conditions, i.e., electricity available from renewable sources. Such a smart and green community of consumers autonomously adapts its energy consumption by enabling a local aggregator to (1) integrate their demand into a common view and, (2) re-schedule the community demand given the renewable energy supply and the consumers’ demand time preferences. In this paper, we evaluate the developed scheduling algorithm using benchmark data to validate our proposal implementation over existent technology.

Supported by Comunidad de Madrid (Spain) under the grant Talent Attraction with reference 2017-T1/TIC- 5184.

1 Introduction

Information and communication technologies (ICT) are essential tools for enabling energy efficiency as well as establishing new energy services and solutions [4]. In particular, the Internet of Things (IoT) [2] and the Smart Grid [3] together can provide the foundational infrastructure and use of advanced information, control and communication technologies to save energy, reduce cost and increase reliability and transparency. Connected devices (e.g, household items, machines, or gadgets) can automatically influence each other so increasing the overall potential for energy savings and the range of management systems’ involvement. ICT can also play an important role in shaping consumer behaviour [5]. Furthermore, the real exploitation of renewable sources for energy supply presents multiple challenges not only for utilities, grid and system operators, but also for the consumers who do not see this type of information on their utility bills [6].

Demand response (DR) programmes appear to bolster not only energy efficiency but also renewable energy resource management initiatives as to handle their sophisticated planning and operation scheduling requirements [1, 7]. In this short paper, we present a cooperative DR system model which is designed to promote behavioural changes in small or large communities of electricity consumers. The community will target common interests (i.e., to be green) that create the need for involved entities to reach binding agreements and coordinated behaviour. We implement and analyse the resource (i.e., the renewable supply) allocation process, initially conceived in a centralised way by means of a data collector called the Aggregator. This entity provides the community with the scheduling of the total demand taking into account both the renewable supply available from the local utility providers and the costumers’ consumption time preferences. Experimentation with estimated values and benchmarks throws feasible performance cost that validates the viability of the system implementation over existent technology (i.e., Z-Wave or IEEE 802.11i standards).

Fig. 1.

A user (or household) equipped with an energy consumption scheduler or home energy manager on a portable device (i.e., an app on a smartphone or tablet) that is connected to a communication network. A community consists of a set of Consumers sharing electricity supplier or substation.

2 System Model

Figure 1 illustrates the main roles and processes within the adopted cooperative demand response framework. In our proposal, Consumers adapt their energy consumption cooperatively on a centralised way; that is, they share their demand schedule with a data collector called Aggregator. The Aggregator facilitates the integration of the energy consumption information and implements an optimised resource allocation algorithm in response to the Utility’s supply conditions, in particular, targeting renewable sources.

2.1 Consumer System Design

Let ${\mathcal {N}}$ denote an ordered set of Consumers that are willing to cooperate in the pursuit of global community targets (i.e., become greener), sending their data to the Aggregator. Each Consumer ${i \in \mathcal {N}}$ has a set of household appliances ${\mathcal {A}_i=\{}$washer, dryer, coffee makers, cooker, fridge, TV, alarm, light controller, water heating, AC system,${\ldots \}}$. Each consumer then pre-allocates¹ a certain amount of fixed demand² as well as variable consumption resulting from their utilisation planned for the upcoming 24 h. The daily fixed demand for consumer ${i \in \mathcal {N}}$ is denoted by ${f\mathcal {D}_i= \sum _{t=0}^{23}\sum _{a_{ij} \in \mathcal {A}_i}fx_{i,a_{ij}}^t}$ as the aggregated load of non-shiftable local consumption of the appliances and frequent behaviours. Variable energy demand is considered flexible since consumer preference for an appliance to start within a particular time interval is also taken into account. For each appliance, there is a execution window (i.e., a closed interval) denoting a minimal starting time and a maximal ending time. In other words, Consumer i will keep/set the following data for his/her appliance $a_{ij} \in \mathcal {A}_i$ as in Table 1.

Table 1. Appliance configuration

2.2 Aggregator System Design

A centralised system with aggregation tasks communicates with the Utility as well as with the Consumers as shown in Fig. 1. An algorithm is originally built to optimise the allocation of the expected electricity supply from renewables amongst the community’s Consumers and according to the their expressed preferences. We denote by ${\mathcal {RW}^t}$ the energy supply generated from a set of renewable sources at a time slot ${t \in \{0,\ldots ,23\}}$. The Aggregator can easily compute the daily fixed demand for the whole community of consumers at a time t as ${f\mathcal {D}^t=\sum _i^N f\mathcal {D}_i^t }$, which should not reach the worst case such that ${\small \sum _i^N \sum _{t=0}^{23} f\mathcal {D}_i^t \gg \sum _{t=0}^{23} \mathcal {RW}^t}$. By contrast, aggregation of the variable consumption is an optimisation problem given the consumers’ time preferences.

The Aggregator will execute a scheduling of the community’s requested variable demand per hour $v\mathcal {D}_i^t$ when the aforementioned worst case does not apply, and aiming at ${\forall t \in \{0,\ldots ,23\}, \sum _i^N (f\mathcal {D}_i^t + v\mathcal {D}_i^t) \le \mathcal {RW}^t}$. We face here a global centralised optimisation problem to which there exists a unique Nash bargaining solution such that: ${\forall i \in \{1\ldots \mathcal {N}\}, \mu _i^t = f\mathcal {D}_i^t + min \{ \mathcal {F}(v\mathcal {D}_i^t) \} \le \mathcal {RW}^t}$, where ${\mathcal {F}(\cdot )}$³ is in charge of shifting the variable demand given Consumers’ appliance preferred activation time. Algorithm 1 displays a round-robin strategy over the matrix of all appliances’ operation preferences and the remnant of the RW supply vector after deducting the total fixed demand (Algorithm 1 – lines 1–3). Upon reaching the optimisation objective, the Aggregator will notify the community that an agreement has been reached and privately release the reallocated demand vector $\overrightarrow{\mu }_i, \forall i \in \mathcal {N}$.

Fig. 2.

Computational cost in Cases 1–4 using SA and strategies RR, randomness and consumers with heterogeneous/homogeneous number of appliances.

3 Algorithm Validation

We evaluate the performance of Algorithm 1 with simulated⁴ data of consumers’ fixed and variable consumption demands at different case scenarios for appliances’ fixed and variable consumption, i.e., Case 1 for high consumptions, Case 2 for high fixed expends, Case 3 when variable is high and, Case 4 for low consumption communities.

Figure 2 illustrates the reallocation processing cost of communities with up to 60 consumers with up to 4 appliances, which is taking 3 min in the worst case. In fact, worst case occurs when appliances demand high variable consumption (Cases 1 and 3) and the algorithm performs a random strategy. On the other hand, we found that the factor incurring the highest performance cost on our algorithm is consumers holding a different number of appliances to schedule; 6 min in the worst case (triangle-up line in Fig. 2) and applying a sequence with the first player being the same every time. Figure 3 throws best outcomes over communities with low variable demand and the same number of appliances per consumer.

Fig. 3.

Performance at the four case scenarios with consumers holding (left) different number of appliances, and (right) the same number of appliances.

4 Conclusions

Smart communities, capable of identifying patterns in energy consumption, will be able to reduce or shift their use of the utility resource, making the overall consumption more sustainable and efficient. Unlike the majority of previous Demand Response strategies that focus on pricing and aim at reducing the energy cost and the peak-to-average ratio, our solution tends to promote a transformation of the whole energy value chain by making consumer communities cooperate targeting the available renewable energy supply. In this paper, we have shown the performance cost of a centralised scheduling algorithm (less than 1 min cost) for different size of communities and consumption patterns. Immediate future work relates to the algorithm testing with real traces from the Birmingham Living Lab and the real implementation of both algorithm and home controllers on a pilot testbed, paying special attention to the system and network security.