Decentralising active network management through distributed constraint optimisation

  • Dimitrios Athanasiadis

Student thesis: Doctoral Thesis

Abstract

Electricity networks are becoming more complex due to the introduction of distributed energy sources. The power grid will face fundamental changes in its structure and behaviour. In addition, new technologies will be required in order to maintain the stability of the network. A key enabling technology for the future power networks is autonomous distribution networks which operate in cooperation with distributed grid intelligence and active network management. Local control can be achieved through fast control and communications and needs to be coordinated with the overall system controls. The primary issues for distribution network operation with a high penetration of distributed resources is power flow management as well as voltage control. Active Network Management can be described as the control and management of generation and load in distribution networks. The main concept is to satisfy network limits such as voltage, power and frequency while at the same time increase the connected generation output with the minimum cost. There is a need for coordination between generators to maintain the balance in the network and avoid investment costs.This thesis considers the introduction of Distributed Constraint Optimisation as a way of providing Active Network Management. It is an agent-based coordination method that is able to coordinate generators’ output without violation of network constraints. This method, which is drawn from the Artificial Intelligence community and was previously used for smaller problems such as meeting scheduling, is studied and evaluated for use in power systems networks in order to provide solutions in a decentralised way.Case studies consider both DC and AC power flow management and the solution of the economic dispatch problem. DC power flow management under DCOP provides optimal solutions for radial networks while AC power flow management is examined from a theoretical standpoint until the limitations of distributed constrained optimisation software are addressed.
Date of Award1 Apr 2015
LanguageEnglish
Awarding Institution
  • University Of Strathclyde
SponsorsEPSRC (Engineering and Physical Sciences Research Council)
SupervisorStephen McArthur (Supervisor) & Ivana Kockar (Supervisor)

Cite this