Remaining lifetime modeling of power transformers: individual assets and fleets

Abstract

Uncertainty analysis is an essential part of models that estimate remaining lifetimes of power transformers. Uncertainty can be included in modeling the insulation aging process. If the uncertainty becomes too large, the insulation condition can, in principle, be determined by measuring the DP value, and more accurate projections can then be made. A drawback of the model presented here is the required input data, i.e., average values of dynamic load and ambient temperature, and maximum values, because the degradation rate is not a linear function of temperature. This information is not available for most transformers. However, even if the loading data are available only over times when the load approaches the rated value, a reliable estimate of the paper degradation can still be made. The observation that tap-changer failure is still the major transformer failure mode supports the simulated results presented above, which show that a transformer failure peak due to paper insulation failure is expected only after several decades of service. Although prediction of transformer performance may be considered speculative because of uncertainties in the model parameters and incomplete service data, it can serve as a tool to compare different maintenance and replacement strategies. Thus annual replacement strategies appear superior to equal loading because the latter provides no guidance on the order of replacement. It has also been shown that transformers should be replaced before the beginning of the failure wave and that variation in annual load growth between transformers leads to spreading of the replacement wave. A disadvantage of a distributed transformer loading is the reduction of the average remaining lifetime due to relatively fast degradation of the more heavily loaded transformers. Comparison of these strategies requires a technical reliability model, which is the foundation of risk-based asset management.