On the hosting capacity of distribution networks for solar power

Sammanfattning: There will be changes in energy production and consumption that will affect the performance of electricity distribution networks. Electric vehicle charging will increase consumption. The installation of solar photovoltaic (PV) units will increase production, change the energy flow and affect the power quality. The installation of solar PV units or electric vehicle (EV) charging has a limit above which may unacceptably affect the normal operation and power quality of the distribution networks. This limit is referred to as the hosting capacity. The hosting capacity at low voltage was found to be most often limited by voltage rise while at medium voltage thermal overload may also be limiting in the distribution networks.Three fundamentally different methods to quantify solar PV hosting capacity for single-phase and three-phase units were investigated:  deterministic, stochastic and time-series. The methods differ in the input data, accuracy, computation time, consideration of uncertainties and time-related influence. The assessment of connection application for screening or detailed analysis is also considered.Solar power production, energy consumption and distribution networks’ all have uncertainties associated with them. It is helpful to distinguish between two types of uncertainties when quantifying the hosting capacity: aleatory (“certain”) and epistemic (“uncertain”) uncertainties. A stochastic approach, ‘mixed aleatory-epistemic’, was applied to a large number of distribution networks to estimate their hosting capacities. The stochastic approach and models were applied also for quantifying the hosting capacity of electric vehicle charging. A deterministic method was applied to determine the hosting capacity considering the thermal overload phenomenon for both PV and EV.The planning risk also referred to as the planning level, has been applied to the stochastic approach. It is applied in form of a percentile value. The planning level entails that a distribution network operator accepts the possibility of exceeding the limit and willing to take it to a certain extent. The hosting capacity for a distribution network is quantified considering the performance index (planning level) and limit. The 90th percentile has been used in the hosting capacity.The time-of-day (ToD) and time-of-year (ToY) concepts were introduced. The time-of-day exemplify the relevant part of the day, and the time-of-year show the parts of the year that are applicable for hosting capacity studies when there is high solar power production. The time-of-day of 10 am to 2 pm has been applied. The period from 21st March to 21st September was the applied time-of-year. The latter two, ToD and ToY, need to be defined for the application of the concept to other areas than those covered in the thesis. It was shown that the hosting capacity would be underestimated by 11% if an incorrect ToD is used.Extensive measurements of voltage magnitude and solar power production, with a 10-minute resolution, were obtained for hundred 10/0.4 kV distribution transformers in Northern Sweden. A method of obtaining the ‘background voltage’ from the measurements was formulated. The background voltage (including its uncertainties) was found to be one of the factors with the greatest effects on the hosting capacity. It is also important to obtain the background voltage during the correct ToD and ToY.Stochastic models for distribution networks were built and the hosting capacity for 1300 low voltage network where studied. The outcome shows that three-phase solar PV units have 3-6 times higher hosting capacity than single-phase. The model and method developed can be used as a planning tool by distribution network operators (DSOs). The inclusion of the uncertainties and correct handling of planning risks is paramount for decision making by DSOs. The results obtained show that details of background voltage variations should be considered from measurements and appropriate ToD/ToY should be used. The quantification of the hosting capacity requires both load and voltage measurements from the distribution networks.The methods with the example illustrations in this thesis are applicable for distribution networks’ hosting capacity quantification. The approaches are a useful tool for planning and decision-making amid uncertainties.