Modelling of Active Distribution Grids for Stability Analysis

Abstract

In the last years the share of distributed generation connected into distribution grids has increased considerably. As their number increases, distribution and transmission network operators are becoming aware on the risks DG can represent on the stable operation of national power systems. To cope with this, the grid code requirements are becoming more and more demanding in order to ensure the secure and reliable supply of energy to the end users. Early grid code requirements were asking Distributed Generation units to disconnect from the grid when disturbances occurred. This was sufficient because the DG penetration was not so high. Nowadays, when most of the European power systems must handle a large share of DG units, more complex and stringent requirements must be fulfilled by the DG units. By reviewing different grid codes this thesis, shows that the technical connection guidelines are varying from country to country. Moreover as the grid connection of DG units with power electronics interface become more widespread, new requirements have found their place in these grid codes. Some of the new capabilities which non−synchronous DG units have to handle during fault occurrence in the grid are: fault ride−through capabilities, reactive current injection or absorption, power oscillations damping and synthetic inertia. In Europe and North America, a regulatory harmonization approach is seen by the introduction of standards and grid codes for large interconnected power systems, for example the ENTSO−E grid code and the IEEE 1547 standard. In Norway investments in small scale hydro units and wind turbines are numerous and expected to increase also as a result of incentives as the green certificates market. Approximatively 98.5 % of the electricity production originates from large hydro power plants. But there is also a large potential for small scale hydro generation. As the distribution grids in the regions where all this potential lies, are in general an ageing infrastructure, the DSOs are concerned about the technical issues they will have to handle to integrate all DG units. The focus of this PhD work is to investigate the technical grid code requirements related to the integration of small scale hydro generators in the future Norwegian active distribution grids. To narrow the research of this work, two main research topics were chosen. I. The first was to investigate the Low Voltage−Fault Ride Through requirement and to identify potential shortcomings of it. Two specific topics were studied: 1. The adequacy of the external power systems modelling for assessing the LV−FRT capabilities of DG units 2. The impact of voltage phase angle variation on the LV−FRT capabilities of DG units II. The second research topic deals with the development of reduced order models of Active Distribution grids (ADG). The aim is to develop practical methods for establishing model equivalents of ADGs that can be applied by TSOs, when performing systems studies. Also here two specific research areas were chosen: 1. The development of dynamic equivalents of ADGs for rotor angle transient stability analysis 2. The development of dynamic equivalents of ADGs for rotor angle small signal stability analysis Within these research areas the PhD work contributes with understanding on how the dynamic equivalents for transient stability can be obtained for a test ADG. Moreover, it was studied how the characteristics of a disturbances impact the identification of coherent generators. For the case of small signal dynamic equivalents the work contributes to the slow coherency theory, related to the identification of slow coherent groups of generators when the modelling of synchronous generators are increased and when the excitation system is included. This research shows that for ADGs with large penetration of small scale hydro units, the oscillation modes similar with the inter−area modes as for TPSs are the inter−machines modes within ADGs. This is because of a good damping of these low frequency oscillation modes. The thesis shows that for ADGs coherency among the DG units some show up within different local plant modes. Linear analysis with simple models for generators fails to identify correctly the groups of coherent DGs as the groups which can be recognized by running a time domain analysis. A method is proposed which uses time domain decomposition of the state variables within inter−machines modes to determine the coherent groups. Having this method available, the computation of phase and magnitude from the time response for a certain oscillation mode can be done easily. For an ADG the complex Euclidean Distance can be then used to cluster the units in coherent groups. The proposed method is later used to obtain the parameters of equivalent groups of coherent generators. In the last part of this work this method is combined with a model parameter identification algorithm to determine the parameter of aggregated generators. The thesis can be summarised as in the following: 1. First, an overview of different national grid codes for DG integration is presented and some topics which were not covered in the LV−FRT requirement were identified and investigated. These topics are the inadequacy of external power system modeling and the absence of voltage phase angle variation. The study of these two research topics (the inadequacy of external power system modeling and the absence of voltage phase angle variation) are representing the main contributions of this PhD research to the LV−FRT requirement. 2. In the second part of the thesis practical methods for establishing model equivalents of ADGs were developed for the purpose of TPS studies. Dynamic equivalents for transient and small signal stability were considered. 2.1 When computing the dynamic equivalents for transient stability it was observed that they are dependent of the characteristics of a disturbance (e.g. location, duration and type. Although these equivalents are disturbance dependent, they provide a good basis for estimation of the Critical Clearing Time (CCT) as well as of the transient stability limits with respect to the original model. 2.2 For the case of dynamic equivalents for small signal stability, the classical method of slow coherency was studied. Further it was shown how the level of modeling accuracy of the synchronous generator and of the excitation system impacts the identification of slow coherent generators. It was shown that the use of linear analysis for simple models of synchronous generators fails to identify correctly the groups of coherent DGs as seen by running a small disturbance time domain analysis. To cope with this problem, a method is proposed which uses the time domain decomposition of state variables to determine the coherent groups and to obtain the parameters of equivalent generators.