Performance Assessment And Design Of Delay-timers In Industrial Alarm Systems

In process industry,alarms are configured to notify operators of equipment malfunctions,process deviations,or any other abnormalities in the plant.However,unsophisticated alarm configuration may result in numerous of false alarm and nuisance alarm or even alarm floods for operators and reduce its performance as a safeguard for plant operations.In the past,many catastrophic incidents happened due to poor performance of alarm systems.Therefore,a scientific and efficient alarm system is crucial to ensure plant safety and uninterrupted operation.To measure the performance of a plant's alarm system,it is necessary to evaluate various performance indices.Motivated by this,this thesis develops quantitative relationship among commonly used alarm techniques and performance indices.In this thesis,a new method is first proposed to compute the false alarm rate(FAR),missed alarm rate(MAR)and expected detection delay(EDD)for alarm design techniques that can be modeled via Markov chains.The proposed method is then applied to deadbands and conventional delay-timer techniques,and advantages are summarized compared with existing algorithms.Generalized delay-timer with more flexibility is a widely used alarm design technique in industries,where instead of consecutive n samples in the conventional case,n1 out of n samples are considered to raise an alarm.For generalized delay-timer,the three important performance indices are calculated via the proposed method,and validated through Monte-Carlo simulations.Finally,alarm system design to meet all the requirements on FAR,MAR and EDD via delay-timer is discussed and industrial data are used to demonstrate the design procedure.