MV Underground Distribution Network Failures and Correlation to Ambient Variables

Electrical utilities are continuously searching for new ways to cut costs and effectively increase the quality of service being supplied to their clients, which is closely related to the number of failures and the time it takes to locate and repair them. A helpful input for this continuing effort would be to have an estimate of the type of failure when it occurs. The analysis presented in this paper is part of an effort being carried out to estimate failure types based on an analysis of observable indirect variables. These indirect variables must be readily available after the failure detection, through a Smart Grid infrastructure, to enable a future application to carry out the estimation. This paper lays the groundwork on ambient and historical variables that display a relationship with various types of MV cable network failures. The paper presents relevant data and analyses the relationship between the various failure types and indirect variables, such as date and time, loading, historical data, daily rainfall, accumulated rainfall, dew point and ambient temperature. Relationships between indirect variables and failure types are established to help identifying the most relevant variables for the design of the estimator.