Alarm management in a LDPE plant in fuction of the operation mode: start-ups and stops

Identifier:  TFG:3376

Handle: https://hdl.handle.net/20.500.11797/TFG3376

Authors:  Toneva, Simona Atanasova

Abstract:
Technology has evolved really fast in the last years, increasing the automation and easy control in the industries, but this evolution is ending up in an overwhelming amount of information on the control panels, that distracts and stresses the operators draining their attention and becomes counter-productive. An alarm is a visual and auditory warning of an abnormal operation condition, a process deviation or a functioning error that needs to be addressed by an operator. It is key to acknowledge that they should not be used as an information source and that if a situation does not require a mitigating action, it is not an alarm. The enormous amount of useless alarms (e.g. alarms that are repetitive and annoying due to fluctuations on the operation conditions, multiple alarms that show the same abnormal situations and become redundant, alarms that do not require a mitigation or are not relevant, etc.) decreases the operators’ attention and make them systematically ignore the alarms due the short period of time in between them. A poorly functioning alarm system has been proved to be a contributing factor to incidents and major accidents, that’s why alarm management is extremely necessary. This project gathers and analyzes the situations that cause most of the alarm floods in a low-density polyethylene plant: the stops and the start-ups, containing hundreds of alarms that are not considered off-normal in those specific cases, like low temperatures and pressures. The alarms are then classified by the equipment or plant zone they belong to. There are lots of methods to solve bad actors (deficient alarms) depending on the problem (intermittences, permanent alarms, floods…) usually based on time delays for activation or deactivation, really useful when talking about the treatment of individual alarms. As this project addresses operation modes that contain large groups of alarms, the main used method consists in the implementation of dynamic alarm suppression. A dynamic alarm suppression mutes a whole group of alarms called targets based on the change of some operation state -or the activation of some alarms- called triggers. This tool is not only useful for stops and start-ups but for every industrial plants that has more than one operation mode: batch plants, plants that change their recipe and readapt the conditions to product, etc. This project considers the configuration of a main dynamic alarm suppression application based on the stop of the plant and other individual dynamics for the equipment which stop and start-up does not necessarily depend on the stopping order of the plant, with the goal to reduce at least half of the alarms present in the stops and start-ups, and improve the plant general monthly KPIs by a 5%.