|
Safety and Integrity
|
|
Alarm Management
|
|
Process Engineering
|
|
Remote Monitoring
|
|
Hydrocarbon Accounting
|
|
Pipeline Leak Detection
|
|
Production Data Management
|
|
Environmental Monitoring
|
|
Alarm Management  Effective warning systems are a vital part of every industrial process.
The integrity of such systems is impaired by standing alarms and nuisance alarms, indeed a system can be rendered completely ineffective by an alarm flood during a process disruption, exactly at a time when the system is needed most.
Most process plants have far too many alarms. This results in numerous standing alarms, repeating and ‘nuisance’ alarms and ‘alarm avalanches’ under plant-upset conditions. Poor alarm performance has been identified as a significant contributory cause of many serious incidents including an explosion at a Texaco refinery in 1994, the Channel Tunnel fire and the Three Mile Island incident.
Methodology
Our alarm rationalisation methodology was developed with Shell Expro as part of that company’s response to the EEMUA guidelines for alarm systems and the IEC-61508 standard for instrumented protective systems. Real Time’s methodology is consistent with both standards and has the added benefit of providing an effective way of dealing with the very large numbers of alarms that often need to be reviewed.
Existing assets
The methodology for existing plants follows seven steps. The first five form a once-off process. The final two should be repeated (at an appropriate frequency), as part of a programme of continuous scrutiny and improvement.
- develop a comprehensive alarm schedule – obtain from DCS configuration data if possible;
- review alarm performance – obtain data from alarm/event logger; identify standing alarms, nuisance alarms, peak and background alarm rates;
- categorise alarms where possible (e.g., trips, pre-alarms, fail to trip on demand);
- review categories, then any exceptions and one-offs – use a risk graph technique similar to IEC-61508 to determine if an alarm is required; if so, what priority; identify required operator action when alarm occurs;
- where practicable, use technical means to minimise standing alarms and alarm floods (e.g., auto-suppression for out-of-service items);
- review alarm performance – obtain data from alarm/event logger; identify standing alarms, nuisance alarms, peak and background alarm rates;
- review alarm limits, deadbands and auto-suppression logic; revise if necessary.
New plant projects
The methodology for new plant projects is similar, with the following exceptions:
- if the alarms have already been defined (e.g., on the P&IDs or in HAZOP recommendations), use this as the alarm schedule;
- if the alarms have not yet been defined, establish a procedure whereby every alarm must be justified, and have a defined operator response.
Once the plant is operational, the final two stages of the ‘existing asset’ process should be undertaken, and repeated at an appropriate frequency thereafter. |