Enhanced local voltage control of active distribution networks with high penetration of PVs

Abstract

The main goal of this thesis is to overcome voltage related problems with increased penetration of photovoltaics (PVs) in the active distribution networks (ADNs). Hence, the thesis aims to develop a novel voltage controller to act in a decentralized local control strategy to regulate voltage, to enhance the hosting capacity of ADNs. Moreover, the developed problem formulation is extended to consider the stochastic nature of PV power output and load demand for higher accuracy of the results. In modern power systems, where a considerable number of converter interfaced generation (CIG) units (e.g. PVs) are increasing, the conventional distribution networks face issues from voltage regulation aspects. The network faces overvoltage problems with increased PV generation in peak generation hours (midday). With increased PV generation in the network, especially at peak periods of generation and low demand, there will be a massive amount of reactive power flow from distribution towards the upstream network. This leads to an increase in grid losses on the network, and also requires greater capacity in system components. ADNs are introduced to contribute to the voltage regulation with the high participation of intermittent renewable energy resources in the system, instead of using conventional grid reinforcement. This thesis proposes a decentralised voltage control strategy to enable a further amount of PVs in the system without exceeding pre-defined limits. The work aims to tackle the problems in previous local voltage control methods in the literature (PF-power and Q-voltage methods). Previous methods are not able to overcome the overvoltage issues and tend to increase the grid losses, especially with the high penetration of PVs in ADNs. Towards addressing the voltage related concerns, a computationally efficient algorithm is developed to minimize reactive power flow through the network electrical bottlenecks. In the proposed approach, all controllers are implemented locally, where the communication devices are eliminated, mitigating the investment cost significantly. Another challenge in the future ADNs is related to the stochastic characteristics of PVs and loads, where, the deterministic approaches may not reach to desired accuracy results with the limited number of time-domain simulations. The current state-of-the-art studies need further consideration for PV power output and load demand when defining the uncertainty. With a scenario generation based technique, the system uncertainties, PV generation output and load demand have been probabilistically modelled and studied with Monte Carlo Simulations. The generation scenarios are reduced to a number of scenarios that is enough to represent the desired variety of samples, and simulations are carried out for these scenarios. To investigate the practicality and scalability of the proposed approach, the proposed methodology has been applied to 33-nodes (radial) and 69-node (meshed and radial configurations) test feeders. The results prove that the developed controller can increase system hosting capacity by about 20% of the systems base power. Furthermore, network active power losses are considerably decreased by 30% concerning existing methods. Moreover, the transformer tap changer experiences fewer tap operations over a day. Thus, feeder hosting capacity can allocate more PVs in the ADN. Besides, the accuracy of results is improved with the stochastic assessment of uncertain values for each generated scenario. Consequently, a more accurate perspective is gained with the overall outcomes for enhancing hosting capacity in ADNs.

Date of Award	31 Dec 2021
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Sinisa Durovic (Supervisor) & Victor Levi (Supervisor)