Real-time thermal state and component loading estimation in active distribution networks

Merkebu Degefa

Research output: ThesisDoctoral ThesisCollection of Articles

Abstract

Highly stochastic loading and distributed generation in the emerging active distribution networks means that electric utilities need to deploy intelligent network management tools in order to use their assets to the fullest. Real-Time Thermal Rating (RTTR) provides the possibility for short term and even real-time active distribution network management, enabling the network to run closer to an overload state without damage. In this dissertation, pertinent developments and proposals are presented in three stages on the path towards the development of a real-time monitoring and operation system for active distribution networks. The first stage is the development of distribution network component thermal models for real time implementation. In this dissertation, a numerical model of the air-gap convective heat transfer for underground cable installations inside unfilled conduit is developed. In addition, a dynamic thermal model is developed for prefabricated secondary substation cabins. The most dominant but difficult to solve heat transfer mechanism, natural convection, is modelled by introducing the stack effect principle into the problem. Measurements from a scaled model of prefabricated substations, measurements from actual cabins and 3D Finite Element Method (FEM) simulations are used to validate the numerical model. In the second stage, this dissertation explores the usability of customer level automatic meter reading (AMR) measurements for distribution network state estimation and for load forecasting applications. A method to forecast substation level loads with their respective confidence intervals using hourly AMR metered customer level consumptions is presented. The forecasting and monitoring of a distribution network in real-time can be met with the modeling of classified type load classes. However, it requires careful incorporation of the necessary factors, such as within-group and between-group correlations of customer classes. Binding the aforementioned findings, in the third stage, a framework for day-ahead hour-by-hour thermal state forecasting and thermal ratings of distribution network components is proposed and studied. This work has demonstrated that up to three hours ahead thermal state forecasting of an active distribution network can be achieved with an acceptable level of accuracy. In this dissertation, the benefits and practical implications of the real-time thermal rating are investigated. The introduction of real-time thermal rating in an active distribution network management system enhances the loading capacity significantly compared to static rating. This has been revealed through an increased utilization of installed DGs and through better integration potential of additional DGs.
Translated title of the contributionReal-time thermal state and component loading estimation in active distribution networks
Original languageEnglish
QualificationDoctor's degree
Awarding Institution
  • Aalto University
Supervisors/Advisors
  • Lehtonen, Matti, Supervising Professor
  • Lehtonen, Matti, Thesis Advisor
Publisher
Print ISBNs978-952-60-6307-2
Electronic ISBNs978-952-60-6308-9
Publication statusPublished - 2015
MoE publication typeG5 Doctoral dissertation (article)

Keywords

  • active network management
  • demand response
  • distributed generation
  • load forecasting
  • natural convection
  • prefabricated substation
  • real-time thermal rating
  • state estimation
  • smart meters

Fingerprint Dive into the research topics of 'Real-time thermal state and component loading estimation in active distribution networks'. Together they form a unique fingerprint.

Cite this