|
Contents
| Quality Encyclopedia
| Discussion Blogs
7.1 Criticality
analysis
Part two of a series.
The following is
an excerpt from The Reliability
Engineering Handbook by Bryan
Dodson and Dennis Nolan,
© Quality Publishing. It may be ordered from the Quality
Publishing Order Form..
The objective of an FMECA
is to identify all failure modes in a system design. Its purpose is to
identify all catastrophic and critical failure probabilities so they can
be minimized as early as possible. Therefore, the FMECA should be started
as soon as preliminary design information is available and extended as
more information becomes available in suspected problem areas.
The effects of all failure
modes are not equal with respect to the total impact on the system concerning
safety and overall system performance. The designer, faced with this dilemma,
needed a tool that would rank the significance of each potential failure
for each component in the system’s design alternatives. Because of
the need for such justification, the criticality analysis function was
added to the Failure Mode and Effects Analysis (FMEA) process, thus creating
Failure Mode, Effects, and Criticality Analysis (FMECA).
This tool has been used
extensively by the military in the last three decades. In recent years,
more commercial industries have been requiring the FMECA to evaluate new
designs and even more recently to improve the reliability of existing
equipment. Military Standard 1629 is a good reference for Failure Mode,
Effects, and Criticality Analysis.
7.1.1 The
qualitative approach to FMECA
This approach should be
used when specific failure rate data is not available. Failure modes identified
by the FMECA process are assessed by their probability of occurrence.
To establish qualitative measures of occurrence, severity, and detection,
criteria must be established that subjectively relate to the overall effect
on the process. Examples are offered in Tables 7.1, 7.2, and 7.3 to serve
as guides in establishing qualitative measures. The product of the measures
of occurrence, severity and detection is called the Risk Priority Number
(RPN). Tables 7.1, 7.2, and 7.3 are for example only. The numbers or criteria
assigned to any particular ranking system are at the discretion of the
user. Detailed instructions on how to use this criterion on the FMECA
form are explained in Section 7.1.4.
7.1.2 FMECA
quantitative approach
Method 102 outlined in
MIL-STD-1629 is the quantitative approach used for the FMECA process.
Figure 7.1 is the worksheet used for this method.
Table
7.1. Occurrence probabilities.
| Rank |
Occurrence
Criteria |
Occurrence
Rates (cycles, hrs, etc.) |
| 1 |
Unlikely.
Unreasonable to expect this failure mode to occur. |
— |
| 2 |
Isolated.
Based on similar designs having a low number of failures. |
1/10,000 |
| 3 |
Sporadic.
Based on similar designs that have experienced occasional failures. |
1/1,000 |
| 4 |
Conceivable.
Based on similar designs that have caused problems. |
1/100 |
| 5 |
Recurrent.
Certain that failures will ensue. |
1/10 |
|
NOTE: The ranking criteria selected must be consistent throughout
the FMECA. |
Table
7.2. Severity Probabilities.
| Rank |
Severity
Criteria |
| 1 |
Minor.
No noticeable effect. Unable to realize that a failure has occurred. |
| 2 |
Marginal.
Annoying. No system degradation. |
| 3 |
Moderate.
Causing dissatisfaction. Some system degradation. |
| 4 |
Critical.
Causing a high degree of dissatisfaction. Loss of system function. |
| 5 |
Catastrophic.
A failure which may cause death or injury. Extended repair outages. |
|
NOTE: The ranking criteria selected must be consistent throughout
the FMECA. |
The failure mode and criticality
number (Cm) is the portion of the criticality number for the item due
to a particular failure mode. This criticality number replaces the RPN
number used in the qualitative method described in the previous section.
The Cm for a failure mode is determined by the expression
Cm = b a l pt
(7.1)
where:
b = conditional probability
of loss of function,
a = failure mode ratio,
l p = part
failure rate, and
t = duration or operating
time.
Table
7.3. Detection probabilities.
| Rank |
Detection
Criteria |
Probability |
| 1 |
Very
high probability of detecting the failure before it occurs. Almost
always preceded by a warning. |
80%–100% |
| 2 |
High
probability of detecting the failure before it occurs. Preceded by
a warning most of the time. |
60%–80% |
| 3 |
Moderate
probability of detecting the failure before it occurs. About a 50%
chance of getting a warning. |
40%–60% |
| 4 |
Low
probability of detecting the failure before it occurs. Always comes
with little or no warning. |
20%–40% |
| 5 |
Remote
probability of detecting the failure before it occurs. Always without
a warning. |
0%–20% |
|
NOTE: The ranking criteria selected must be consistent throughout
the FMECA. |
The b values represent
the analyst’s judgment as to the conditional probability that the
loss will occur and should be quantified in general accordance with the
following:
| Failure
effect |
Probability
of loss of function |
| Actual
loss |
b
=1 |
| Probable
loss |
0.10<b
<1.00 |
| Possible
loss |
0<b
<0.10 |
| No
effect |
b
=0 |
The failure mode ratio,
a , is the probability that the part or item will fail. If all potential
failure modes of a particular part or item are listed, the sum of the
a values for that part or item will equal one. Individual failure mode
multipliers may be derived from failure rate source data or from test
and operational data. If failure mode data are not available, the a values
should represent the analyst’s judgment based upon an analysis of
the item’s functions.
Part failure rates, l
p, are derived from appropriate reliability prediction methods
using mean-time-between-failure (MTBF) data or possibly other data obtained
from handbooks or reference material. Manufacturers often supply failure
data; however, it is important that the environment the item will be subjected
to is similar to the environment the manufacturer used when obtaining
the failure data.
The operating time, t,
is usually expressed in hours or the number of operating cycles of the
item being analyzed.
The a and b values are
often subjective, thus making the supposed quantitative method somewhat
qualitative. All things considered, it is generally understood that the
FMECA process is a qualitative method of analysis.
|
