PPCL was recently implement in a Saudi refinery with a reported 85% reduction in standing alarms and a 50% reduction in alarms per minute.Here are the details. There were a total of 993 process variables that both had process variable alarm limits provided and data present in the plant historian. Nearly 200 other process variable tags were identified in the preliminary stage that may not be getting recorded properly in the historian, these have been noted.
The calculated alarm count over the time period (ie alarms on the alarm list display) had an average of 36 process variables in alarm.
New limits or maintenance were recommended for 54 process variables. These changes give an 85% reduction in standing alarms and a 50% reduction in annunciation rate. The revised limits sit on the boundary of normal process operation. This both reduces nuisance alarms and enhances operator trust in the remaining alarms.
For the process portion of the gas processing plant examined in the second week, 445 tags with alarm limits were considered. The historic alarm counts for this period calculated from historic data were 20-44 standing alarms with an average of 29 at any time.
The alarm limits were revised by identifying tags where the current limits were inconsistent with the current operation. This included cases where variables were often in alarm as current bad actors, known and unknown process or mechanical faults, and alarm limits so far outside the range of normal operation as to provide late or no warning of process upset. Fifteen tags were identified for maintenance attention, limits for 13 tags were modified, and 42 limits were identified for removal. The performance afterward had 0-12 standing alarms with 2 average for a 90% reduction in alarm count and a 66% reduction in annunciation.