The impact of high-voltage circuit breaker condition on power system reliability

Abstract

The electric power grid is a complex, ageing system that requires substantial investment to meet societal needs. Reliability analysis tools are required to assist in decision-making concerning system development and asset management. Power system reliability analysis typically models components with non-varying failure rates, but failure rates can increase due to wear-out. Component reliability models have been developed for transmission lines and transformers. However, models for high-voltage circuit breakers (HVCBs) are also necessary as they are essential for grid protection. This research proposes a methodology for incorporating HVCB condition-monitoring data into power system reliability analysis. The condition monitoring data is used for fault diagnosis and health prognosis of the HVCBs. Survival analysis techniques are used to investigate time-to-event data and construct a component reliability model. Finally, analytical methods for reliability analysis are used to quantify the system-level reliability indices. The research findings were presented in three separate publications: 1. A health index model based on failure modes, effects, and criticality analysis (FMECA) was proposed for HVCBs. The health index model was based on two HVCB failure modes: the HVCB tripping without a command and failing to trip on command. These two failure modes form the main aspects of protection system reliability that are affected by an HVCB’s technical condition. The failure mechanisms that lead to each failure mode were identified and graded based on their severity, occurrence, and detectability to produce a risk priority number (RPN). The RPN was then used to justify the contribution of each failure mechanism towards the health index. Condition monitoring data was used to assess the amount of degradation of each failure mechanism. The trip coil current (TCC) was used to evaluate the onset of relevant failure mechanisms due to its availability and capabil ity of identifying failure mechanisms in both the electrical and mechanical sections of an HVCB. The health index was calculated using a weighted sum average. A case study revealed no direct correlation between health index and component age, emphasising the need for methodologies that can quantify component health based on condition data. 2. The Cox proportional hazards model (PHM) was utilised to convert the HVCB health indices into failure rates. The Cox PHM comprises a baseline failure rate, which is dependent on time and a link function, which is dependent on covariates and regression coefficients. The regression coefficients for each covariate of the Cox PHM, such as voltage, manufacturer, substation, and age, were calculated using an outage database containing information on planned and unplanned outages. The baseline failure rate was calculated using Breslow’s method. The health indices were then used in the link function instead of the regression coefficients and covariates while maintaining the baseline failure rate to generate a condition-dependent failure rate. An Exponential and a Weibull model were fitted to the HVCB uptime vs event data to calibrate the condition-dependent failure rate model. The Brier score indicated that the Exponential model was the best fit to describe HVCB survival times. Therefore, the Exponential model was used to calibrate the calculated condition-dependent failure rates to historical records. 3. The impact of an HVCB’s technical condition on the system-level reliability indices was assessed using approximate methods. The condition data of the HVCBs was used to calculate two condition-dependent reliability parameters. One parameter, denoted as λs, was related to the rate at which an HVCB trips without a command, while the other parameter, denoted as Pm, was associated with the probability of an HVCB failing to trip on command. These condition-dependent reliability parameters were then incorporated into existing methodologies that can evaluate the impact of protection system failures on system reliability. A sensitivity analysis revealed that the reliability indices were most affected by the condition-dependent reliability parameter Pm, primarily due to the common-mode failures it induces. This research developed a comprehensive methodology for utilising HVCB condition data at the component level and analysing its effects on the system-level reliability indices.