Our client operates a Floating Production, Storage and Offloading Facility (FPSO) in the North Sea. Their alarm control system was poorly designed and they realised they had a problem with nuisance alarms.
They asked ABB Consulting to carry out an alarm management health check. The health check confirmed that they had a number of issues and in particular suffered from an overload of alarms. To help the client solve their alarm problems ABB began an alarm rationalisation programme.
We began the project by carrying out an alarm analysis. Using the data sent from the alarm system offshore, all the alarms were identified and ranked in order of frequency to highlight the worst nuisance alarms.
A little way into the project, data was sent from a different alarm system console offshore. As the alarms had been considered a nuisance, the sounder on the console had been switched off so they did not sound and had been so for almost 10 years. The alarms from this console had originally been excluded from the rationalisation process by the client. After some investigation, ABB found that that they were in fact genuine alarms mostly caused by failing gas detectors. In 1 month over 412,000 gas detector alarms were generated. The personnel on the asset believed the detector alarm only indicated a calibration (drift) error. In fact, the alarm also indicated a complete failure of the detector. Figure 1 below shows the extent of gas detector faults and results of the first remediation visit carried out by the F&G vender.
The fire and gas vendor was called out to investigate the situation and found that the detectors causing the nuisance alarms had multiple faults, and worryingly, almost 20% of the detectors checked out were totally dead. All faulty and dead detectors were replaced to bring the system back to full functionality.
As the gas detector alarms were being ignored (and in some cases there was a 2oo2 voting system), in the event of a genuine gas release where one detector in a 2oo2 voting configuration was dead, the GPA would not have been activated. In one zone, both detectors in a 2oo2 configuration were found to be dead. This could have led to a serious incident and possible loss of life. An internal safety notice detailing the findings was sent to the operator’s other assets. This revealed that another asset had the same problem.
As well as the 412,000 silent gas detector alarms; in the same month, 78,000 HVAC alarms had also gone unheard. Investigations instigated by ABB found these to be caused by pressure control dampers being set to manual, (alarms were generated through pressure differentials when doors were opened), meaning the HVAC system was not operating as per design to provide the required positive pressure to the protected areas.
At its peak; not counting process alarms, the combined average gas detector and HVAC alarm rate was 1 every 2 seconds. Once all the remediation work had been carried out; the gas detector and HVAC alarm rate dropped significantly to 4 or 5 a day, as shown below, and the console was returned to normal operation, annunciating alarms to the Control Room Operators.
This serious problem had been misunderstood and ignored for almost 10 years, and if not for the work of ABB, would have remained a hidden risk.
The gas detector and HVAC alarms were a key layer of protection. When holes in the temporary refuge on the asset had been discovered, an Operational Risk Assessment (ORA) had been carried out. This erroneously concluded that it was safe to carry on production due to the protection afforded by the gas detection and HVAC systems. Had this situation gone unnoticed and the HSE had uncovered the problem, (and they had already expressed concern over the asset’s process alarm rate), there is a distinct possibility that the HSE may have served notification to cease production due to the safety risk to personnel.