The three pre steps of FMEA Analysis are explained below:

Step 1  -  Severity

In this step, all failure modes associated with product functioning and their effects are determined.  Corrosion, deformation, and electrical short-circuits are some examples of failure modes.  Since failure modes lead to a chain process, it is necessary to list each failure mode according to its function and in technical terms.  Then failure effect for each failure node is analyzed.  Noise, injury to user, and degraded performance are some examples of failure effects.  A Severity Number (S) is assigned to each failure effect ranging from 1 (no danger) to 10 (critical).  This helps in prioritization of failure modes and effects.  If a failure effect has a severity rating of 9 or 10, then it would cause injury to user or end up in litigation.  Such failure modes are immediately eliminated and the design is changed.

Step 2  -  Occurrence

In this step, the causes of each failure and its frequency are identified and documented in technical terms.  An earlier documentation for similar processes would be of value here.  Examples of failure causes are excessive voltage, improper operating conditions, erroneous algorithms, etc.  Failure causes are weaknesses in design.  An Occurrence Ranking (O) is assigned to each failure mode, ranging from 1 to 10.  It can also be expressed in %.  Non-safety failure modes have ranking >4, if their Severity Number are 9 or 10, they have ranking >1.  All these cases demand determination of action.  Occurrence Ranking is based of the product and specifications of customer.  This step is known as the detailed development category of FMEA.

Step 3  -  Deduction

In this step, the actions determined are tested for their efficiency.  Design verification is done, and proper inspection methods are chosen.  To do this, the engineer should look at the current system controls and assess their ability to prevent failure modes or detect them before it moves over to customer.  The testing, monitoring, analysis, and other techniques used in similar system controls to detect failures can be identified.  A Detection Number (D) is assigned to the inspections or planned tests according to their ability to detect or prevent failures.  D measures the extent of risk of a failure escaping detection.  Thus, a higher ranking of D indicates the chance of failure detection is low.

continue reading part 3

After the above steps, Risk Priority Numbers (RPN) are computed.  Once the S, O, and D are ranked, then RPN is arrived at by multiplying the three numbers:  RPN = S x O x D.

RPN pinpoints the areas that are of greatest concern.  RPN should be done for the entire life-cycle, so that failure modes with highest RPN are set right.  There are some failure modes which may be lesser in severity, but greater in frequency and less detectable.  After allocation of values, the next step is to note the recommended actions with targets and dates of implementation.  These actions may include testing, specific inspection, quality procedures, redesign, limiting environmental stresses, etc.

After implementation, it is necessary to check the new RPN for confirming the improvements.  The tests can be noted in a graphical form.  In a word, in the face of failure modes, you have to first eliminate the failure mode, then minimize its severity.  Then take steps to reduce its frequency and increase its detection possibility.

The use of FMEA is limited by the experience of committee members regarding previous failures.  In the event they cannot identify the failure mode, then they have to approach consultants for solving this.  In FMEA, documentation is an important factor for implementation.  If FMEA is used as a top-down tool, it may identify only major failure modes in a system, when compared to Fault Tree Analysis (FTA).

It can complement FTA when used as a bottom-up tool.  FMEA does not have the capacity to identify complex failure modes which involve multiple failures within a subsystem.  It also cannot predict failure intervals of a particular failure mode up the upper level system/subsystem.

There is also a risk that the multiplication of S, O, and D rankings may end up in rank reversals, so that a less severe failure mode is shown as a more severe one.  The reason is that the rankings are numbers based on ordinal scale, which only compare one ranking to another, but does not measure the degree of difference.

Many specific national and international standards require that FMEA should be used in the evaluation of product integrity.

You may be interested in: