Distributed Energy Storage Control for Dynamic Load Impact Mitigation
Maximilian Zangs, Peter Adams, Timur Yunusov, William Holderbaum, Ben Potter
The future uptake of electric vehicles (EV) in low-voltage distribution networks can cause increased voltage violations and thermal overloading of network assets, especially in networks with limited headroom at times of high or peak demand. To address this problem, this paper proposes a distributed battery energy storage solution, controlled using an additive increase multiplicative decrease (AIMD) algorithm. The improved algorithm (AIMD+) uses local bus voltage measurements and a reference
... age threshold to determine the additive increase parameter and to control the charging, as well as discharging rate of the battery. The used voltage threshold is dependent on the network topology and is calculated using power flow analysis tools, with peak demand equally allocated amongst all loads. Simulations were performed on the IEEE LV European Test feeder and a number of real U.K. suburban power distribution network models, together with European demand data and a realistic electric vehicle charging model. The performance of the standard AIMD algorithm with a fixed voltage threshold and the proposed AIMD+ algorithm with the reference voltage profile are compared. Results show that, compared to the standard AIMD case, the proposed AIMD+ algorithm further improves the network's voltage profiles, reduces thermal overload occurrences and ensures a more equal battery utilisation. Energies 2016, 9, 647 2 of 20 Demand Side Management (DSM) strategies for Distributed Energy Resources (DER), aim to alleviate the impacts of PHEV/EV home-charging and are a favoured solution. developed a distributed DSM algorithm that implicitly controls the operation of loads, based on game theory and the network operator's ability to dynamically adjust energy prices. Focusing on financial incentive-driven DSM strategies, in , a Time-Of-Use (TOU) tariff and real-time load management strategy was proposed, where disruptive charging is avoided by allocating higher prices to times of peak demand. Financial incentives have also become a drive towards optimising the operation of Battery Energy Storage Solutions (BESS) and Distributed Generation (DG) when including PHEV/EV into the problem formulation  . Research focused on grid support has been driven by the need to deliver long-term savings and to avoid the immediate costs and disruption of network reinforcements and upgrades. This area of research proposes the implementation of alternative solutions to support the adoption of low carbon technologies, such as EVs, heat pumps and the electrification of consumer products. To reduce the resulting increased peak demand, Mohsenian-Rad et al. developed an approach of direct interaction between grid and consumer to achieve valley-filling, by means of dynamic game theory  . In , a Multi-Agent System (MAS) was used to manage flexible loads for the minimisation of cost in a dynamic game. The use of aggregators has been proposed to allow the participation of a number of small providers to participate in network support, such as grid frequency response    . Yet without the availability of power demand forecasts, real-time control needs to be implemented. Real-time DSM can either be implemented in a centralised or distributed control approach. In the former, a central controller relays control signals to its aggregated DERs, whereas the latter allows each DER to control itself. A common form of controlling DERs in this mode of operation is set-point control  . Using set-point control on multiple identically-configured DERs would yield optimal operation conditions if each DER's control parameters (e.g., bus voltage) were shared. In a system without sharing network information, DER control algorithms have to be improved to prevent, for example, devices located furthest from the substation from being used more frequently than others. This paper therefore presents an individualised BESS control algorithm that lets distributed batteries respond to fluctuations in real-time local bus voltage readings. The proposed algorithm is based on the robust Additive Increase Multiplicative Decrease (AIMD) type algorithm, yet implements a set-point adjustment based on the location of the controlled BESS. It will be shown how these home-connected batteries can mitigate the impact of additional loads (i.e., EV uptake), whilst assuring that all BESS are cycled equally. The key contribution of this work can be summarised as a novel distributed battery storage algorithm for mitigating the negative impact of dynamic load uptake on the low-voltage network. This algorithm uses an individualised set-point control to regulate bi-directional battery power flow and, for convergence, extends the traditional AIMD algorithm. As a result, the developed battery control method reduces voltage deviation, over-currents and the inequality of battery usage. Reducing this usage inequality leads to a homogeneous usage of all of the distributed batteries and, hence, prevents unequal degradation rates and unfair device utilisation. The remainder of this paper is organised as follows: Section 2 gives some background to related work on AIMD algorithms on which this research is based. Section 3 outlines the EV, network and storage models used in the research. Additionally, it explains the assumptions that accommodate and justify these models. Section 4 elaborates on the proposed AIMD control algorithm (AIMD+). Next, Section 5 details the implementation and scenarios used for a set of test cases. For later comparison, this section also outlines a set of comparison metrics. Section 6 presents and discusses the results, followed by the conclusion in Section 7.