Optimisation of smart-and Vehicle-to-Grid charging strategies in distribution networks

Abstract

Currently in the energy system of the Netherlands, lower level consumer demand flexibility is rather obsolete due to sufficient capacity limits of the distribution network and supply of capacity mechanisms by large industrial actors. However, in the upcoming 10- 20 years, the power system is in transition to become decentralised with a higher share of renewable energy sources and significant increase in consumption. An operational control structure in the power system, where private consumers provide flexible capacity, is an effective and economical efficient approach to make sure the regulated process of electricity generation to supply at consumers is secure and reliable. Currently as a result of the EU Energy Efficiency Directive of 2012, an institutional base is presented for development initiatives of demand response in Europe [17]. Technical and regulatory standards now enable demand response flexibility to be offered on the wholesale and retail energy market and allow for consumer participation [34]. Demand response schemes are usually distinguished by the various motivation methods offered to the participating consumers. Programs include in general two control methods, centralised direct load control or time-based and incentive-based DR. Because these schemes rely on demand response decision-making by means of a centralised (multi) aggregator perspective, direct load control can be precisely adjusted to technical (local) grid constraints [69]. Practically, the objective of DR in this research is used to reduce congestion in distribution grids by moving part of BEV energy demand from (evening) peaks to the afternoon or night with direct control. By achieving these measures potential benefits arise, including the most profound in the distribution grid [29]:

• Optimising local grid assets by increasing the utilisation factor, and thereby maximise asset efficiency and subsequently decrease costs, which is beneficial for the DSO
• Scheduling of peak charging demand to aid congestion in distribution grids.

The modelling of the demand response charging strategies in Amsterdam fills the knowledge gap towards handling congestion for the DSO. It also provides a new study that addresses the potential to postpone future distribution grid investments by using charging strategies specifically for Amsterdam. The main research question that this study addresses is therefore:

What is the value of demand response management in a Vehicle-to-Grid network and does it provide increased benefits to smart charging for consumers and the distribution system operator in Amsterdam?

In order to grasp the subject of congestion prevention within the time limits of graduation, the scope of this study is limited to assess the first mentioned item by modelling charging demand, and subsequently simulate optimal demand response charging strategies for a case study in Amsterdam’s local power grid. The motivation for this study is threefold. Firstly, providing insight and recommendations in Amsterdam’s BEV charging demand to compute the potential to provide demand flexibility by making use of a large-scale charging sessions data (2017-2018). Literature that uses such a vast data set for Amsterdam is scarce. Therefore, typical demand behaviour for daily and seasonal variation or periodicity is assessed by using a method of time series analysis and local regression analysis to derive time- and load-flexibility parameters. These parameters de- scribe the measures for which demand response is generally defined. Secondly, a linear programming model is developed that optimises charging demand of smart charging and V2G charging strategies for demand response purposes to aid congestion at local feeders in the distribution grid of Liander. Lastly, an information-task exchange protocol is de- scribed from the perspective of a multi-aggregator to perform coordinated DR.

Further investigation about the usable amount of flexibility is simulated with linear programming for charging strategies (smart charging and V2G charging) to aid in local feeder congestion. In this research a case study is performed on data of a low-voltage feeder under different congestion situations, and a range of different number of BEVS connected to explore future distribution grid implications. Results show that the model’s performance works especially well during peak demand periods throughout the day. In addition, the V2G strategy outperforms charged prices per kWh in almost every simulation compared to the uncontrolled scenario, by both charging during periods of low prices and discharging during periods of high prices even though the LP model does not optimise on prices. Thus, the V2G strategy allows the DSO to postpone grid investments in a number of cases while simultaneously the consumer almost always receives remuneration for its delivered services. The V2G strategy therefore provides a significantly added value over a smart charging strategy, by allowing electric vehicles to be charged whenever the feeder exhibits congestion. Future work includes analysis towards the value of DR in a liberalised system regarding the subject of the split-incentives challenge. Who initiates demand response (consumer, retailer, aggregator, DSO) and how should the benefits be divided along the supply chain. Handling an optimisation problem of DR for holds strong requirements for an global system balance in which neither participating actors are discriminated and the whole system benefits. An assessment of relational dependencies between participators in DR should be incorporated in designing charging strategies.