## Beta 1 |

Title | Condition Management |

Author | Jensen, Rikke Helbirk |

Supervisor | Holbøll, Joachim (Electric Power Engineering, Department of Electrical Engineering, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark) Madsen, Henrik (Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark) Jensen, Lars Bai (DONG Energi S&D) |

Institution | Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark |

Thesis level | Master's thesis |

Year | 2008 |

Abstract | Today the error statistics in the electrical industry consists of a statement of the numbers of failures per component per year, or concerning cables, as the number of failures per 100 km. Even though it is known that the risk of failure for a certain component for example is affected by the operational and ambient conditions it is exposed to, then these parameters are not included in the error statistics. This master thesis investigates the opportunities of using known available parameters for estimating the risk of failure for power switches and cables, including cable joints, in the 10 kV power grid. An analysis of the data material shows that a statistical valid model cannot be set up for power switches. This implies that only cables and cable joints will be investigated. The statistical method used in the analysis is the generalized linear model. The generalized linear model identifies which parameters are significantly correlated with the risk of failure of the different components. A significant correlation implies that the risk of failure for the individual components can be estimated. The result of the analysis shows, that the significant parameters for cable joints are; type, age and whether the joint are connecting two different or the same type of cables. The result of the analysis for cable joints shows, that if the 10% of the cable joints, which are estimated as the one with highest risk of failure, are collected, then 50% of the cable joints, which had an error in the period March 2008 - September 2008, are included in this group. Equivalently it is shown that the significant parameters for cables are; type, age, length and distance to the main station. It is found that by grouping the cables, which are estimated to have the 10% highest risk of failure, then 42% of these cables, which had an error, are included in this group of the 10%. In a similar investigation, based on the error statistics reported to DEFU, it is found that only 13% of the cables, which had an error, are included in the group with the highest 10% of risk of failure. It can be concluded that the generalized linear model is not only usable but also a better tool than the traditionally methods. Furthermore it is summarized, that other parameters would be interesting to include in the model, but since the data is not available for all components, including the non failed ones, it is not possible right know. Recommendations for further collecting of data and analysis are to be given. |

Admin | Creation date: 2009-05-06 Update date: 2009-05-06 Source: dtu ID: 242450 Original MXD |