8.2.3 FAILURE MODE EFFECT ANALYSIS (FMEA)

К оглавлению
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 
34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 
51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 
68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 
102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 
119 120 121 122 123 

Failure Mode Effect Analysis (FMEA), sometimes called Failure Mode, Effects,

and Criticality Analysis (FMECA) is intended to result in preventive actions.

FMEA and FMECA are ‘‘before-the-fact,’’ exercises and you will most

likely use this technique when evaluating technical, rather than programmatic,

aspects of the system. A comprehensive FMEA will take a significant amount of

time so be sure to allocate the time necessary.

An FMEA is performed from the bottom up by evaluating the lowest part of

the system first. This involves the Lowest Replaceable Unit (LRU), the smallest

or the least significant unit. In electronic systems, this ‘‘unit’’ is frequently a

resistor or transistor or some such unit. After each unit is reviewed, it is used

as a ‘‘building block’’ to ultimately re-create the entire system. The usual way

to conduct an FMEA is to look at a ‘‘failed’’ unit and making a prediction by

asking the question: ‘‘What happens when this unit fails?’’

The purpose of the FMEA exercise is to identify critical components in a

system and evaluate the impact of the failure of that component and then, if

warranted, provide an alternate path, or back up or change the unit so that the

impact of failure is removed. By way of example, we did an FMEA on two

alternative data links from Cape Canaveral to Houston back before the first

mission was run from Houston. The system was designed to have two completely

diverse data paths. One route was completely land-line (telephone line)

and the other was completely RF (microwave). We traced the design through

the entire thousand miles from the Cape to the Mission Control Center (MSC)

in Houston only to find that both diverse routes entered the Telco building just

outside the Control Center, and both routes went through the same amplifier.

Consider the effect of that failure!

Using reliability data is a good way to predict failure rates for components.

For instance, if ‘‘X’’ fails, it will take the ‘‘Y’’ function with it. However, the

probability of ‘‘X’’ not failing is 0.999999, so it’s not likely ‘‘X’’ or ‘‘Y’’ will be a

big contributor to overall system failure.

There are two kinds of failure, total and partial. In many systems, you will

get a different result with a partial failure than with a total failure. Other than

the obvious, a partial failure may affect the productivity of another unit in the

system and that unit will change the direction of the failure. For instance,

changing the bias on an electronic circuit will cause a different effect in a secondary

circuit than if the circuit were completely dead.

Software to provide FMEA analyses is available as ‘‘Relex FMEA/FMECA’’

from:

Relex Corporation

40 Pellis Road

Greensburg, PA 15601

Phone: 724-836-8800

Fax: 724-836-8844

For additional information on FMEA, see:

McDermott, Robin E., et al. The Basics of FMEA. Portland, Ore.: Productivity

Press, Inc., 1996.

Stamatis, Dean H. Failure Mode and Effect Analysis: FMEA from Theory to

Execution. Milwaukee, Wis.: ASQ Quality Press, 1995.