Page 32 - Root Cause Failure Analysis

Root Cause Failure Analysis Methodology 23

No matter what the event is (i.e., equipment failure, environmental release, accident,
etc.), the evaluation must quantify all the variables associated with the event. These
data should include the operating setup; product variables, such as viscosity, density,
flow rates, and so forth; and the ambient environment. If available, the data also
should include any predictive-maintenance data associated with the event.

Who Was Involved? The investigation should identify all personnel involved,
directly or indirectly, in the event. Failures and events often result from human error
or inadequate skills. However, remember that the purpose of the investigation is to
resolve the problem, not to place blame.

All comments or statements derived during this part of the investigation should be
impersonal and totally objective. All references to personnel directly involved in the
incident should be assigned a code number or other identi$er, such as Operator A or
Maintenance Craftsman B. This approach helps reduce fear of punishment for those
directly involved in the incident. In addition, it reduces prejudice or preconceived
opinions about individuals within the organization.

Why Did It Happen? If the preceding questions are fully answered, it may be pos-
sible to resolve the incident with no further investigation. However, exercise caution
to ensure that the real problem has been identified. It is too easy to address the symp-
toms or perceptions without a full analysis.
At this point, generate a list of what may have contributed to the reported problem.
The list should include all factors, both real and assumed. This step is critical to the
process. In many cases, a number of factors, many of them trivial, combine to cause a
serious problem.

All assumptions included in this list of possible causes should be clearly noted, as
should the causes that are proven. A sequence-of-events analysis provides a means for
separating fact from fiction during the analysis process.

What Is the Impact? The evaluation should quantify the impact of the event before
embarking on a full RCFA. Again, not all events, even some that are repetitive, war-
rant a full analysis. This part of the. investigation process should be as factual as possi-
ble. Even though all the details are unavailable at this point, attempt to assess the real
or potential impact of the event.

Will It Happen Again? If the preliminary interview determines that the event is
nonrecurring, the process may be discontinued at this point. However, a thorough
review of the historical records associated with the machine or system involved in the
incident should be conducted before making this decision. Make sure that it truly is a
nonrecurring event before discontinuing the evaluation.

All reported events should be recorded and the files maintained for future reference.
For incidents found to be nonrecumng, a file should be established that retains all the