Distribution Transformer Risk Modelling
Zoya Gorman
Creation of a Failure Model for Distribution Transformers, that outputs probability of failure and allows identification of Risk Levels and Mitigation techniques.
Distribution transformers generally last reliably for decades, however, when they do occasionally fail, they can cause significant outage costs and consequence to a utility. Quantifying reliability of assets is a challenging task due to lack of manufacturer reliability testing, varying operating duties, maintenance and often replacement before failure. Many current guidelines being used are prevention techniques through replacement of the asset, without any accurate calculation of when the asset would fail. Consequently, development of a failure model which can factor in different variables related to the asset and can quantify the risk of failure will be extremely beneficial for risk management and expenditure planning.
This project created a Failure Model for the fleet of distribution transformers at United Energy. The model takes on the form of a Cox Model with a Weibull distribution as the baseline hazard function. A Cox model allows analyse of the effect of different variables on failures, in addition to calculation of probability of failures. The variables analysed include Age, Cyclic Rating and Overloading on transformers. Through this model, risk levels are identified and mitigation techniques are provided.
Page Views:
Distribution transformers generally last reliably for decades, however, when they do occasionally fail, they can cause significant outage costs and consequence to a utility. Quantifying reliability of assets is a challenging task due to lack of manufacturer reliability testing, varying operating duties, maintenance and often replacement before failure. Many current guidelines being used are prevention techniques through replacement of the asset, without any accurate calculation of when the asset would fail. Consequently, development of a failure model which can factor in different variables related to the asset and can quantify the risk of failure will be extremely beneficial for risk management and expenditure planning.
This project created a Failure Model for the fleet of distribution transformers at United Energy. The model takes on the form of a Cox Model with a Weibull distribution as the baseline hazard function. A Cox model allows analyse of the effect of different variables on failures, in addition to calculation of probability of failures. The baseline hazard function is a function of time, and analyses the effect of Age, while the variables analysed in the Cox Model are Cyclic Rating and Overloading on transformers. The final model predicts annual failures within 5% of historical failures.
Through this model, risk levels are identified, and mitigation techniques are provided. Risk levels included an increased risk for Overloaded transformers. Overloaded transformers are 2.6 times as likely to fail than Underloaded transformers, and overloading by greater than 15% results in a significant increase in risk. Four zone substations where transformers are at higher risk of failure were identified, along with an increased risk in early-life of transformers. Identification of all these risk levels were made possible through the Cox Failure Model created and as a result, United Energy can make changes to risk management and expenditure planning to reduce transformer failures and consequences suffered from failures.
Industry Partner: United Energy
