Control of Grid Interactive PV Inverters for High Penetration in Low Voltage Distribution Networks

摘要

Regarding of high density deployment of PV installations in electricity grids, new technical challenges such as voltage rise, thermal loading of network components, voltage unbalance, harmonic interaction and fault current contributions are being added to tasks list of distribution system operators (DSOs) in order to maintain at least the same power quality as before PVs were not revealed. Potential problems caused by high amount of PV installations can be avoided with technical study of both power system and power electronics areas that also benefit for new grid connection requirements. Any network management scheme or weakly-prepared grid connection requirements without paying attention to PV integration problems can bring potential risk of unintentional disconnections of these generating plants that likely increase payback time and extra energy losses of these renewable energy sources. On the other hand, unnecessarily strict grid connection requirements may cause less utilization of solar potential and may lead to additional cost on PV plants.PV based generating plants are basically interfaced to electricity grid via power inverters. Hardware and control design requirements of these inverters may depend on grid connection rules which are forced by DSOs. Minimum requirement expected from PV inverters is to transfer maximum power by taking direct current (DC) form from PV modules and release it into AC grid and also continuously keep the inverters synchronized to the grid even under distorted conditions. Chapter 2, therefore; overviews the latest ancillary services such as real power reduction during over frequency/over voltage events and reactive power control for static grid voltage support function of PV inverters. In case of high density of PV integration, grid connection rules which do not consider interaction among multiple PV inverters with the grid may allow unlimited number of PV plant connections and lead to unstable network operation. Harmonic emissions from multiple inverters connected to the same feeder and resulting in a network resonance can be a good example for this problem but PV inverters connected to highly capacitive networks are able to employ extra current and voltage harmonics compensation to avoid triggering network resonances at low order frequencies. The barriers such as harmonics interaction, flicker, fault current contribution and dc current injections from inverters can be figured out as long as the maximum PV hosting capacity of networks fully exploits available solar potential of geographical region. Therefore, grid voltage rise and thermal limits of network components will be considered as the most prevalent barriers in the thesis.One of the focuses in this thesis is to develop a simulation tool and methodology for the estimation of maximum PV hosting capacity of LV distribution networks based on grid voltage rise and transformer hot-spot temperature limitations (Chapter 3 and 4). Rooftop PV installations in power capacity below 6 kWp have widespread usage in residential areas and are usually single-phase connected to 230/400-V grid. Since realistic assessment of PV integration should include both single- and three-phase PV connections, a three-phase load flow script which is able to allow more precise estimation of PV hosting capacity in unbalanced cases has been developed in Chapter 3 for future studies. Modeling of power system components has been revised in threephase coordinates. The developed script has been validated with comparison results obtained from IEEE distribution test networks and from commercial software within the tolerable errors.Current status on planning and operation of distribution networks has been also briefly summarized from PV integration perspective in Chapter 4. The possible network problems arisen from high penetration of PV plants or in other words, network limitation factors against increasing PV penetration to further levels can differ depending on network structures. For example, voltage rise will likely be essential limiting factor in the networks which have long-distant feeders. Unfortunately, there is no a unique network structure, but at least, critical reference LV networks can be characterized from statistical analysis of real networks. Therefore, reference models of critical LV networks (suburban and farm) from literature have been revised and accordingly, their PV hosting capacities have been estimated by means of the developed load flow calculations. Invoking thermal model of distribution transformers can also contribute on more accurate estimation of PV hosting capacity.As the other focus, the most considerable work of this thesis has been dedicated to the local voltage support methods of PV inverters in Chapter 5. The objective here is to compensate voltage rise owing to PV systems by absorbing reactive power from the grid. Thus, more PV power can be allowed for grid connection as long as steady-state grid voltage is in admissible range. However, grid voltage support of PV inverters by reactive power control is limited in distribution networks. The main reasons are high R/X ratio of LV networks, PV inverter current limitation, transformer and cable/line thermal limits with increased reactive power flow. Therefore, the highest voltage drop should be realized with minimum reactive power absorption from the grid. Weak points of voltage support strategies which were already imposed by grid codes have been underlined and two new methods have been proposed. In order to prevent unnecessary reactive power absorption from the grid during admissible voltage range or to increase reactive power contribution from the inverters during grid overvoltage condition, the proposed methods have been inherited from standard cos(φ) and Q(U) methods by combining their properties. Finally, both simulation and experimental validation of these methods have been provided in Chapter 5.