PhD Thesis, Department of Energy Technology, Aalborg University, Aalborg Universitetsforlag, 2016 // 31 January 2016
The increasing share of Distributed Generators (DGs) poses concerns about network stability and their fair participation to the electricity market. Their aggregation in Virtual Power Plants, sometimes in combination with load prosumers, could represent an answer to these issues. Nevertheless Technical Virtual Power Plant still offer limited ancillary services, as reactive power management and spinning reserve. Focus of this thesis is on the control of DGs for providing further and higher quality ancillary services. In particular LV dispatchable DGs, hereinafter called flexible DGs, as stationary battery units (SBUs), microturbines (MTs) and Diesel Engines (DEs) appear promising solutions. The models of a SBU derived from a commercial Electric Vehicle battery and a MT are proposed. Both units are Voltage Source Converter (VSC) interfaced and, following recent Power Electronics state of the art, they are equipped with an LCL output filter. The thesis starts deepening the design of current control scheme for a stable operation in grid connected mode. Such flexible units offer on the other hand the possibility to survive a LV feeder in the future case of intentional islanding or for black start service. Therefore the design of the voltage regulator is developed proposing the compensation of the voltage drop on to grid side inductance. Moreover considering the hierarchical control a comparison between direct and reverse droop configurations is proposed. The participation to local reactive power market implies the precise control of reactive power through DGs. While reactive power regulator tuning is straightforward in case of current controlled VSCs, it is more challenging in case of synchronous generator interfaced DGs, as DEs and small CHP units, due to the high system order introduced considering the electrical machines, real exciter and network equations. The analytical tuning of the same regulator is here developed starting from small signal analyses of the plant, with an approach fitting to any reactance to resistance ratio of the equivalent Thevenin model regarded for representing the network. Such time consuming tuning procedure is then compared with an automated regulator design based on multiple time domain runs driven by a Simplex Optimisation. In particular some novel objective functions relying on typical step response indexes are proposed and assessed.