Energy storage and electric vehicles as a means of mitigating uncertainty in urban microgrids

Abstract

The United Kingdom (UK) government intends to end the sale of new conventional petrol and diesel cars by 2040, and Electric Vehicles (EVs) could emerge as the replacement. This is likely to increase the load on electrical distribution networks, while uncontrolled EV charging could increase load forecast uncertainty. Utilising sufficient Energy Storage System (ESS) power to maintain the networks within their power flow and voltage limits without needing to reinforce the network, while not over using the storage despite the uncertainty, remains a challenge. Similarly, the EVs themselves have been suggested as a flexible load however realising this flexibility also remains a challenge. This Thesis researches the ability of ESSs and EVs to mitigate load and generation uncertainty within urban microgrids. Initially, the technical and economic impacts of uncontrolled EV charging on distribution networks is investigated by combining an extensive real world dataset of EV charging events and domestic household load. It is found that distribution transformer power flow limits will be the first operational limit to be breached when EV penetration reaches 40%. The resulting reinforcement cost that Ofgem would allow Distribution Network Operators (DNOs) to recover from consumers is estimated at £60.81bn - £74.27bn up to 2040. A methodology is then proposed to forecast future uncontrolled EV charging load based on the ‘here and now’ load experienced on the network. In addition, a methodology is proposed to aggregate a number of smart charging EVs to form a Virtual Energy Storage System (VESS) able to deliver services to the distribution network with a high degree of controllability (~99%), while also guaranteeing the energy required by the EVs for their primary purpose of transportation. The VESS is combined with other forms of flexibility to deliver an Enhanced Frequency Response (EFR) service where a fuzzy logic control methodology is proposed to maximise power availability. Finally, a Robust Optimisation (RO) formulation is developed that balances the trade-off between the cost of protecting network operational limits from load and generation uncertainty, against the cost of failing to protect network operational limits. RO requires a linear representation of the power system, and the errors introduced through linearization via sensitivity factors are calculated as up to 1.6% when there is no load and generation uncertainty, and up to 4.0% when there is load and generation uncertainty.