RealityCharting e-book .pdf - SERC Home Page

Paradigm Shift
In addition to using causal factors analysis, the Deepwater Horizon
Accident Investigation Report presented by British Petroleum uses
storytelling, barrier analysis, and a crude form of fault tree analysis to
understand the event. The following is an example of how they used
storytelling to analyze one of the reasons why the blowout preventer did
not seal the well and oil was released.
“Solenoid Valve 103 Condition: During the yellow pod test
performed by Transocean and Cameron after the accident,
both coils on solenoid valve 103 failed to energize, suggesting
electrical faults. The investigation team found no evidence
that this failure occurred after the accident; rather, the team
concluded that this failure condition very likely existed prior
to the accident. (Refer to 5.1 Maintenance of this analysis) A
faulty solenoid valve 103 would mean that the yellow pod
could not have performed the Automatic Mode Function (to
isolate the well), as no pilot signal could have been sent to the
pilot-operated control valve to activate the high-pressure Blind
Shear Ram close function. As described in 5.5 Monitoring and
Diagnostic Capability of this analysis, the rig’s Blowout Preventer
control diagnostic system should have been capable of remotely
detecting the faulty solenoid valve and recording it on the system
event logger.”
Furthermore, Section 5.1 Maintenance of this analysis provides the
following story:
“In September 2009, A BP rig audit team conducted an audit
of Deepwater Horizon. This audit included the maintenance
management system for the Blowout Preventer. One
finding was, “Overdue maintenance in excess of 30 days was
considered excessive, totaling 390 jobs and 3,545 man hours.
Many of the overdue routines were high priority.” This audit,
which the team performed at the end of the rig out-of-service
period for ten-year maintenance and inspection, identified
thirty-one findings that were related to the well control
system maintenance. Of these, six findings related to Blowout
Preventer maintenance; all findings were outstanding as of
December 2009.
22