Representing an alarm system for processes with variables with mixture probability distribution

Alarm systems play an important role in ensuring safety, and preventing event occurrence in industrial plants. One of the most important steps in alarm system designing is estimation of the proper probability density function (pdf). Conventional methods in alarm system designing like, dead-band and delay timers cannot be more effective in case of alarm variable with mixture pdf. This paper presents a new method to design an univariate alarm system with mixture pdf in alarm variables. In this paper three alarm performance indecis are derived for variables with gussian pdf. In proposed method, rasing and clearing alarms are based on the probability values corresponding to the instantaneous alarm variable values in the normal and abnormal pdfs (normal and abnormal reference models). The effectiveness of the proposed method is shown during some simulation and industrial case studies and its performance compared with Reset scenario in delay timers. In one of the case studies, the performance of the proposed method in the DAMADICS benchmark actuators has been investigated.