Reliable and selectable islanding detection techniques for low inertia power system

  • Kianoush Naraghipour

Student thesis: Doctoral Thesis

Abstract

With the development of inverter-based distributed generator (DG) connected to thedistribution network, utilities are becoming concerned about the risk of unintentionalislanding formation.Unfortunately, many existing methods are unable to detect islanding for loads that closelymatch the DG capacity, particularly for higher loading quality factors. Other methods, onthe other hand, reduce the power quality. The complexity and cost of the other methods arealso disadvantages, making them unsuitable.This thesis proposes three innovative methods for islanding detection. They are classifiedin the active methods for equipping inverter-based DG in connection to the grid along withthe local loads. They are based on Q-f droop curve method, voltage amplitude variation(VAV) method and hybrid method of Q-f and VAV methods respectively.The operation of these proposed methods is investigated and simulated for detection ofunintentional islanding when the grid is disconnected. Also, their performance aresimulated for maintaining stable operation during the grid connection, even with the loadswitching.Q-f droop curve method has a good performance with zero NDZ. Also, VAV method hasa fast detection time, however, there is small NDZ for VAV method. To compensate forthe disadvantages of the VAV method, a hybrid method with a short detection time isdeveloped.The operation of the proposed methods, when the DG is generating only active power orgeneration simultaneously active and reactive power for both balanced and unbalancedloads, are simulated. All aforementioned cases are simulated to validate three proposedmethods have the effective operation for both outside and inside the non-detection zone(NDZ). Due to the importance of microgrids in the advancement of decentralised electricitygeneration, three proposed methods are considered to prove their robustness undermultiple-DGs connection.As realistic loads, such as motors, which constitute a substantial portion of the distributionsystem, the performance of the proposed islanding detection methods are simulated with amotor load as well. It is confirmed that the proposed methods can recognise betweenislanding detection and motor starting to avoid any unnecessary disconnection from thegrid.The proposed methods are supported by theoretical, simulation, and Q-f droop curvemethod is validated by experimentation.
Date of Award16 May 2022
Original languageEnglish
Awarding Institution
  • University Of Strathclyde
SponsorsEPSRC (Engineering and Physical Sciences Research Council)
SupervisorKhaled Ahmed (Supervisor) & Campbell Booth (Supervisor)

Cite this

'