Effect of size on electrical performance RN Hampton * , A Smedberg, DF Wald Borealis AB, Stenungsund, Sweden * now at GA Tech NEETRAC, 62 Lake Mirror Road, Bldg. 3, Forest Park, GA 30297 Nigel.Hampton@neetrac.gatech.edu Abstract The electrical breakdown performance, either unaged or after ageing (laboratory or service), is often used as the basis for qualification of a device, design or material. Many of the features that affect these performance levels have been discussed in other documents; contaminants, propensity for water treeing, insulating and semiconducting materials. However the size of cable tested is rarely discussed. This is somewhat surprising as it has been long recognized that electrical failure is an Extreme Value (the Weibull distribution is a member of this family) or weakest link process. In Extreme Value processes the performance of the whole device is determined by the single “weakest link”. Thus when more “weak links” are present the chance of failure is consequently higher: the measured performance depends on weak link concentration or size of the device. Additionally at some dimensions the thickness of the dielectric can influence the breakdown mechanism itself; especially if the thermal influences are present This paper will attempt to discuss a number of these size related issues for both AC & Impulse conditions; these will include: • The effect of the dielectric volume actual mechanism of failure • Prediction of performance on service length cables from short length laboratory tests. This has practical relevance on the selection of appropriate qualification levels which will have direct relevance to service performance. • The requirements for cable quality when increasing the size (thickness or length) installed. 1,0 Introduction Recent predictions show that the world will require 60% more energy by the year 2030. This presents electric power distributors with a very real challenge: “How to maintain the necessary pace of network development and ensure consistently high system performance and reliability”. Reliability will become increasingly more important as regulatory frameworks raise expectations in respect of ‘supply quality’. The transmission and distribution of electrical energy requires efficient and reliable networks with low losses. Traditionally this role has been fulfilled by overhead networks, which have been considered as a lower cost alternative to underground cable solutions [1, 2]. However, with progressive advances in technology, calculations show that the costs of overhead lines and cables are much closer when compared on the basis of ‘Total Cost’. This comparison goes beyond installation costs only and takes into account a broader range of criteria, including fault rates and dielectric losses. Time (Years) World Energy Demand (rel to 100 in 1990) 2030 2020 2010 2000 1990 200 175 150 125 100 75 50 Source Nuclear Oil Renew ables Coal Gas Figure 1 Estimated world energy demand (International Energy Agency) It is now understood that cables are valuable because they are both invisible and reliable. If the cables were not reliable then the necessary repairs would make them visible, as well as costly. The conclusion to be drawn is that reliability equals value and anything that compromises reliability is a “false economy”. This is of vital importance to customers and a commercial imperative for grid owners as transmission reliability gives maximum earnings through volume of transmitted energy. 2.0 Theoretical Basis Failure data can be analysed using a wide variety of techniques; Gaussian, Non Parametric to Extreme Value. The most common approach is to use the Weibull distribution (Equation 1) [3]. Equation 1 shows the most general or 3 Parameter form; however it is quite usual to see the 2 Parameter form where S L =0. β α − − − = ) exp( 1 ) ( L S S S P Equation 1 Where P is the probability of failure at the applied stress S; S is the stress (voltage, electrical stress, number of cycles) applied to the system; S L is the location or threshold parameter, the probability of failure is vanishingly small below this value; α is the magnitude estimator and is referred to as the scale parameter;  β is the mechanism estimator and is referred to as the shape parameter.