Active Management of PV-Rich Low Voltage Networks

This thesis was submitted to The University of Manchester for the degree of Doctor of Philosophy in the Faculty of Science and Engineering.The increased penetration of residential-scale photovoltaic (PV) systems in European-style low voltage (LV) networks (i.e., long feeders with high number of connected customers) is leading to technical issues such as voltage rise and thermal overload of the most expensive network assets (i.e., transformer, cables). As these issues significantly limit the ability of LV networks to accommodate higher PV penetrations, Distribution Network Operators (DNOs) are required to proceed with expensive and time-consuming investments in order to reinforce or replace these assets. In contrast to this traditional approach of network reinforcement, which potentially leads to massive capital expenditure, the transition towards active LV networks where controllable elements, existing (i.e., PV systems) and likely to be adopted (i.e., battery energy storage systems, LV on-load tap changer transformers), can be managed in real-time, poses an attractive alternative. Although several active network management schemes have been recently proposed to increase the hosting capacity of PV-rich LV networks, they are mostly based on managing voltage issues only; and, in general, aim to solve technical issues separately. Integrated solutions aiming at managing simultaneously voltage and thermal issues are required, as recent studies demonstrate that both issues can coexist in PV-rich LV networks. More importantly the majority of studies, which commonly neglect the characteristics of real LV networks (e.g., unbalanced, three-phase, radial, multiple feeders with several branches, different types of customers), use complex optimisation techniques that require expensive communication infrastructure and extensive or full network observability (currently not available in LV networks). However, considering the extensiveness of LV networks around the world, practical, cost-effective and scalable solutions that use limited and already available information are more likely to be adopted by the industry. Considering the above gaps in the literature, this Thesis contributes by proposing innovative and scalable active network management schemes that use limited network monitoring and communication infrastructure to actively manage (1) Residential-scale PV systems, (2) Residential-scale Battery Energy Storage (BES) systems and (3) LV on-load tap changer (OLTC)-fitted transformers. The adoption of the proposed active network management schemes, which makes use of already available devices, information and requires limited monitoring (i.e., secondary distribution substation), allows making the transition towards active LV networks more practical and cost-effective. In addition, to tackle the challenges related to this research (i.e., lack of realistic LV network modelling with high resolution time-series analyses), this Thesis, being part of the industrial project Active Management of LV Networks (funded by EDF R&D) and having access to French data, contributes by considering a fully modelled typical real residential French LV network (three-phase four-wire) with different characteristics and number of customers. Moreover, realistic (1-min resolution) daily time-series household (from real smart meter data) and PV generation profiles are considered while a stochastic approach (i.e., Monte Carlo) is adopted to cater for the uncertainties related to household demand as well as PV generation and location