Emerging power system designs are driven towards supporting the diversity of generation supply through increased flexibility and progressive employment of renewable resources, in order to fulfil sustainable and low carbon energy development strategies. The resulting impact on system response, especially during and following disturbances, can incur negative power system stability issues. With particular attention given to the anticipated reduction and variation of system inertia, concerns are growing with respect to power system frequency stability following large disturbances. This is attributed to the potential involvement of acceleration and magnification in frequency excursions due to reduced inertia and differences in generator (particularly converter-interfaced source) control and responses to disturbances.Despite the fact that new techniques, such as those that produce “synthetic” inertial responses, are evolving to compensate for the absence of inherent inertia from renewable resources, their practical integration into the existing grid is limited at the present stage. This is attributed to the uncertainties underlying when and how these techniques should be applied, with a major barrier to their deployment being a lack of prevailing system inertia information. As the exact amount of real-time response required cannot be known with a high degree of confidence, a risk of “over-responding” would be incurred if inertial responses are deployed widely throughout the system. Moreover, the suitability of adopting present-day frequency-based protection settings in conjunction with aforementioned techniques in future power networks has not been fully investigated, where frequency stability margins, necessitating active regulation, could dynamically vary.As such, the work reported in this thesis focuses on evaluating the impact of variable inertia on the performance of existing frequency-based protections and the feasibility of introducing adaptive solutions as a flexible and reliable approach to improve the performance of affected frequency protection schemes, thereby enhancing future system frequency performance. There are two major contributions in this thesis. Firstly, a Switching Markov Gaussian Model (SMGM) has been proposed with which the real-time inertia estimates can be profiled from observed frequency variations during normal system operation. An optimised error of lower than 10% (taken for a system of an overall inertia equal to 3 seconds) was produced for 95% of the daily estimation if being calibrated with the equivalent inertia derived from generation dispatch data on a half-hourly basis and its robustness can be maintained for a period of up to two hours when losing frequency observations. Secondly, a zonal adaptive Distributed Generation (DG) anti-islanding protection scheme has been developed and demonstrated with protection settings being adjusted in response to estimated levels of system inertia. Enhancement of the performance of DG anti-islanding protection has been tested and demonstrated on a reduced GB power network model. Validity and robustness have been analysed, along with discussions of configuration adjustments and practicalities of adopting reliable adaptive protection schemes.
|Date of Award||1 Oct 2014|
- University Of Strathclyde
|Supervisor||Graeme Burt (Supervisor) & Campbell Booth (Supervisor)|