Design of a prognostic alert system for coal power plants using FMEA and the proportional hazards model
Merve Kaya, Huseyin Carpanali, Ahmet Yucekaya, Mustafa Hekimoglu, Şeyma KayaPurpose
The power plants face failures during start-up or operation which may cause long disruptions in electricity generation, a serious problem for both electricity producer and market operater. The capacity of coal plant is submitted to the day-ahead market for the next day. When the plant is scheduled for the operation, a failure prevents power generation and causes capacity loss and downtime cost. Hence, a framework needs to be developed to estimate the failures before the operation and market settlement.
Design/methodology/approach
In this work, we use historical data for plant failures and analyze the failure types, modes and components using descriptive statistics. Then Failure Mode Effect Analysis is used to analyze the faults, errors and failures for each plant. To address varying risk levels, risk priority numbers are calculated for each failure type that can cause crucial failures for each plant. Then, the Andersen–Gill’s extended proportional hazards model (PHM) is applied to calculate the probability of failures in power plants at different times.
Findings
Using a threshold level for survival probability, which indicates the probability of a failure for each plant, a prognostic alert system is created that can predict the timing of an upcoming failure. The proposed methodology provides both plant and previous failure-based survival curves to help decision makers with the operations.
Originality/value
Coal plants are still crucial for the sustainable operation of electricity markets. The proposed methodology is a novel, generic and promising solution for coal plants as the decision makers will be able to schedule a maintenance before failure using this methodology.