Citation & Copyright (to be inserted by the publisher ) Dielectric breakdown statistics of polyethylene for progressively- censored data M. Carmo Lana and J. Marat-Mendes * Departamento de CiŒncia dos Materiais, Secªo de Materiais Electroactivos (CENIMAT), Faculdade de CiŒncias e Tecnologia, Universidade Nova de Lisboa, 2825-114 Caparica, Portugal Keywords : power cables, dielectric breakdown, Weibull function, polyethylene. Abstract. The dielectric breakdown of thin films of low-density polyethylene (LDPE) electrically aged in an aqueous solution of NaCl under an AC electric field was investigated. A two-parameter Weibull function was used for the dielectric breakdown time to failure. The probability of failure for a sample was obtained by the White method for progressively censored data. Samples aged at different temperatures were compared. The results show that initially the samples aged at lower temperature (25 o C) are more prone to fail, while those aged at higher temperature (50 o C) fail at longer times. This was attributed to a competition between oxidation and diffusion. Introduction Polyethylene is widely used as the polymeric insulation for medium and high voltage power cables. The combined effect of the electrical field and environmental conditions originate electrical aging of the polyethylene [1]. In the long run aging reduces the useful lifetime of cable when dielectric breakdown occurs. Many parameters influence breakdown (temperature, electric field amplitude and frequency, mechanical stress and the type and concentration of ionic species present). The published studies include cables aged under service conditions but it is often more convenient to accelerate aging in the laboratory [1]. Dielectric breakdown is a localized and stochastic process and a statistical approach is often applied. The most common is to fit the data (failure times) using a two-parameter Weibull function [1,2]. Laboratory accelerated electrical aging experiments do not necessarily end in the failure of all samples (some experiments are aimed for studying other aspects of aging prior to failure and quite often samples are aged for specific time periods). However it is unavoidable that some of the samples will breakdown before the predicted aging time is reached. In this situation the set of samples aged are composed of a group of samples that failed prematurely and another group that has been aged for different aging times without failing. If a statistical approach is used in such a set, the data is called progressively-censored (meaning that from a total of n samples only r have suffered breakdown while n r were aged for a given time). The two parameters in the Weibull function are the shape parameter (a) and the characteristic time ( W ) and the probability to failure ( F P ), at a given time t, is given by: () = a W F t t P exp 1 (1) Information about the dielectric breakdown mechanism can be obtained from these two parameters. Both parameters can give some information on the breakdown mechanisms. The methods used for the estimation of parameters are the least squares linear regression method and the maximum likelihood estimator [3]. Several techniques were compared for estimation accuracy, especially bias in the estimated parameters [4,5]. For progressively censored data the White method [6] was recommended and simplified by Montanari et al. for electrical aging [5]. The White method is a weighted version of the least squares linear regression method. * corresponding author, jnm@fct.unl.pt Materials Science Forum Vols. 455-456 (2004) pp. 602-605 online at http://www.scientific.net © (2004) Trans Tech Publications, Switzerland All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of the publisher: Trans Tech Publications Ltd, Switzerland, www.ttp.net . (ID: 193.136.121.72-03/11/06,18:08:31)