Estimation of Alarm Deadbands

Abstract Alarm deadbands in industrial control systems are commonly used to reduce the amount of alarm chatter. While easy to implement, guidelines for setting the deadband value based on the process measurement signal and noise do not exist. Shown in this paper is an application of time series analysis to determine both measurement and time alarm deadbands. It is shown that, similar to Kalman filter design, the optimal value of the deadband is determined by both the measurement noise and the trajectory of the signal immediately preceding the alarm.