Sec 48 Reliability Concepts and Data Analysis

RELIABILITY CONCEPTS AND DATA ANALYSIS
1
INTRODUCTION
mers expect a purchased product to meet or exceed life expectations and to be saf
providing early signals and information on root causes of failure.
ng of early-production units in the field, analysis of warranty data, and systemat
Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 2

INTRODUCTION
ty evaluations often present a challenge beyond that normally encountered in qual
is sometimes referred to as “burn-in.”

iability?

INTRODUCTION
bility data:
le system.
non repairable components within a repairable system).
sential that detailed component-level information on cause of failure be obtained.
ed to mere assessment of overall system reliability. Subsequent subsections descr

INTRODUCTION
lure times, are known. The most common reason for censoring is the need to analyze
n that unit would have failed. For example, if a unit were removed from the field

LIFE DATA MODELS
ility density function (pdf), a survival function (sf), or a hazard function (hf). T
rated (for one possible time-to-failure distribution) in Figure on slide 8.
purpose or another.
can be interpreted as the proportion of units in the population (or from some st

LIFE DATA MODELS
dt. Thus, for a positive random variable, F(t) =
of as a smoothed histogram of a large number of observed failure times. The hf (al

LIFE DATA MODELS

LIFE DATA MODELS
T is a measure of central tendency or “average”
ilure may differ appreciably from other measures of central tendency like the med

ANALYSIS OF CENSORED LIFE DATA
e underlying distribution) plotted on special probability paper. Such plots also
times, hazard function values, and appropriate confidence intervals on some or all
oring for failures, and all censoring at the end of the observation period (typica
mps by the number of tied failures divided by n). This estimate requires no assump

ANALYSIS OF CENSORED LIFE DATA
ting a life distribution, using an assumed model, and estimating various propertie
s about the shape of F(t) outside the range of 30 to 90 thousand cycles. Moreover,

ACCELERATED LIFE TEST MODELS AND DATA ANALYSIS
ut failure for years, decades, or longer. Thus, we might expect that few units will
ars. For such applications, Accelerated Life Tests (ALTs) are used widely, particul
tests for components, materials, or products with a single failure mode.

365 times each day, we could reduce the median lifetime (for many potential failur
g in eventual weakening) of an adhesive mechanical bond or the growth of a conduc
a high level of stress will generally fail more rapidly than it would have faile

capture and describe the dominant aspect of the process. For example, failure may
ernative. However, such a model might provide an excellent fit to the available da

SYSTEM RELIABILITY CONCEPTS
ction of time in operation t (or other measure of use), the operating environment(s
ture to compute system reliability. Complicated system structures can generally be

REPAIRABLE SYSTEM DATA
tances involving nonrepairable components, the assumption of independent and iden
r repair.
e of system repair times for similar systems.
of the system at the time just before the repair and the nature of the repair. Th

OTHER TOPICS IN RELIABILITY
assure up front that the product/system reliability will meet customer requireme
er criteria. The goal of FMEA/FMECA is to identify and possibly remove possible fa
product or a system. It might be done initially for individual subsystems. Then t

ures using a “top-down” approach. First, one or more critical “top events” (such as
ts at a higher level are caused by lower-level “primary events” (e.g., failure of an
ee analysis, however, differs in its basic approach to system reliability. Reliabil
other contributing events, at all levels within the scope of the analysis. The str

rmation for assessing and improving reliability. These include life tests and acce
acturing conditions or operating/environmental conditions that might be encounter

n or a specified percentile of the life distribution of the product. For example,
ment of Defense (1960), Quality Control and Reliability Handbook H-108, provides fa

defects. After all or most of these defective units have failed, population failur
bility. Manufacturers, in their quality and reliability improvement programs, striv

ocess, components such as integrated circuits may be run at high levels of temper
combinations of temperature cycling, physical vibration, and perhaps stressful op
re also expensive and may not be totally effective. By improving reliability throu

d implement fixes to affect all future manufactured units. Thus, later-generation
nic systems depend heavily on software for important functionality and flexibilit

ity differs from hardware reliability in at least one important way. Hardware fai
icular system state or level of system load is reached. The state of the software
not have occurred if the software had been designed to anticipate the conditions

ystem or the application software may correct such problems. In safety-critical sy
and amount of testing or use to which the system has been exposed. In an attempt

Sec 48 Reliability Concepts and Data Analysis

Uploaded by

Copyright:

Available Formats

Sec 48 Reliability Concepts and Data Analysis

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sec 48 Reliability Concepts and Data Analysis

Uploaded by

Copyright:

Available Formats

RELIABILITY CONCEPTS AND DATA ANALYSIS

providing early signals and information on root causes of failure.

ng of early-production units in the field, analysis of warranty data, and systemat

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 2

ty evaluations often present a challenge beyond that normally encountered in qual

is sometimes referred to as “burn-in.”

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 3

non repairable components within a repairable system).

sential that detailed component-level information on cause of failure be obtained.

ed to mere assessment of overall system reliability. Subsequent subsections descr

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 4

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 5

rated (for one possible time-to-failure distribution) in Figure on slide 8.

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 6

dt. Thus, for a positive random variable, F(t) =

of as a smoothed histogram of a large number of observed failure times. The hf (al

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 7

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 8

T is a measure of central tendency or “average”

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 9

e underlying distribution) plotted on special probability paper. Such plots also

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 10

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 11

tests for components, materials, or products with a single failure mode.

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 12

g in eventual weakening) of an adhesive mechanical bond or the growth of a conduc

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 13

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 14

ture to compute system reliability. Complicated system structures can generally be

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 15

tances involving nonrepairable components, the assumption of independent and iden

e of system repair times for similar systems.

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 16

er criteria. The goal of FMEA/FMECA is to identify and possibly remove possible fa

product or a system. It might be done initially for individual subsystems. Then t

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 17

ts at a higher level are caused by lower-level “primary events” (e.g., failure of an

ee analysis, however, differs in its basic approach to system reliability. Reliabil

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 18

acturing conditions or operating/environmental conditions that might be encounter

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 19

n or a specified percentile of the life distribution of the product. For example,

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 20

bility. Manufacturers, in their quality and reliability improvement programs, striv

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 21

combinations of temperature cycling, physical vibration, and perhaps stressful op

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 22

d implement fixes to affect all future manufactured units. Thus, later-generation

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 23

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 24

ystem or the application software may correct such problems. In safety-critical sy

Sec 48 RELIABILITY CONCEPTS AND DATA ANALYSIS 25

You might also like