Enhanced unscented transform method for probabilistic load flow studies

Oke, Oluwabukola A. (2013) Enhanced unscented transform method for probabilistic load flow studies. PhD thesis, University of Nottingham.

PDF - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
Download (3MB) | Preview


The advent of deregulated electricity and the call for sustainable energy practices are major drivers for the continued increase of renewable energy systems within the modern day power network. Dominant among them is the wind energy system whose output is uncertain because of its dependence on the prevailing climatic conditions. This increases the level of uncertainty witnessed within the power system as such, as the penetration of renewable energy systems continue to increase, their effects cannot be trivialised.

Probabilistic load flow (PLF) is employed by power system analysts to account for the effect of uncertainty within the power network. The common technique which is based on Monte Carlo Simulation (MCS), though accurate is very time consuming and for large systems it becomes unwieldy. Alternative approaches with the advantages of the MCS method but with reduced computational burden are required. A viable alternative method should therefore require minimum computational time and burden, be able to accurately model various network uncertainties, be applicable to practical small and large systems, be able to account for the effect of dependency among network variables and possess good overall accuracy.

This thesis proposes a novel approximate approach referred to as the enhanced unscented transform method to meet the requirements of PLF. The method combines the Gaussian quadrature method and the Stieljes procedure with dimension reduction technique in deciding estimation points while the Cholesky decomposition is incorporated to account for the effect of dependency. The performance of the proposed technique is demonstrated using modified IEEE 6, 14, and 118 test systems and a practical distribution test system all incorporating wind farms. Results obtained for numerous scenarios show a good match between the proposed method and the MCS method but with significant computational burden saving. The performance of the method is also shown to compare favourably with other existing PLF methods.

Item Type: Thesis (University of Nottingham only) (PhD)
Supervisors: Thomas, D.W.P.
Keywords: probabilistic load flow, unscented transform, univariate dimension reduction, wind, Weibull distribution
Subjects: T Technology > TK Electrical engineering. Electronics Nuclear engineering
Faculties/Schools: UK Campuses > Faculty of Engineering > Department of Electrical and Electronic Engineering
Item ID: 14040
Depositing User: EP, Services
Date Deposited: 24 Feb 2015 12:31
Last Modified: 16 Dec 2017 15:54
URI: https://eprints.nottingham.ac.uk/id/eprint/14040

Actions (Archive Staff Only)

Edit View Edit View