Application to FMEA Priority Assessments - Probabilistic Composition of Preferences, Theory and Applications - page 111

11.2 FMEA

FMEA is a tool to improve reliability. The basic objective of FMEA is to eliminate
potential causes of failures before they take place.
FMEA may be applied in the development of a product as well as in the
modeling of a manufacturing process. It is preferably applied before beginning
manufacture or assembly, but can be applied to processes in progress, on a cyclic
movement to raise quality. Once developed a product or process whatsoever, it
must be regularly reviewed, its output compared to that imagined, to the attention to
the expected faults added the focus on those really happening in the day-to-day
process management and product use.
Application of FMEA involves recording the possible failures and evaluating its
severity, frequency and detectability objectively in a document, the form of FMEA.
In the form of FMEA are initially recorded functions and main features of the
product or process, modes of potential failures for each function; causes and effects
of each mode of failure and current controls.
FMEA starts basically with the formation of a group of people to identify the
product or process in question, its functions, the types of failures that can occur, the
effects and the possible causes of each failure, to develop the form of FMEA. Then
the risks for each cause of failure are evaluated by means of the three indices. Based
on this assessment, are proposed actions able to reduce these risks by eliminating
the most important causes of failure, those whose combined priority index exceeds a
predetermined threshold, increasing in this way the reliability of the product or
process.
At the stage of risk assessment, evaluations are recorded in the form of indices of
severity (S), occurrence (O) and detectability (D) for each mode of failure, fol-
lowing previously de first of these indices measures the severity

of the consequences attributed to an undetected occurrence of the failure. The
second measures the probability of occurrence, i.e., the frequency of observation
expected for the failure. The third measures the dif

ned patterns. The

culty of detecting the failure in
a timely manner. These indices are used in the traditional approach to calculate the
RPN by multiplying their three values.
The identi

cation of the values for these indices is in principle performed in
meetings of the group of analysts. The development of communication tools in
recent years favors, however, the use of isolated assessments that can be revised
iteratively.
When the group is deciding on values according to one factor, the other indices
cannot be taken into consideration, i.e., the evaluation of each index is independent.
For example, when evaluating the level of severity of a particular effect, a low value
should not be assigned to this index only because the probability of detection is
high.
Further details and reviews of FMEA can be found in AIAG ( 2008 )or
McDermott et al. ( 1996 ), for instance.