IEC 61164:2004

INTERNATIONAL IEC
STANDARD 61164
Second edition
2004-03
Reliability growth –
Statistical test and estimation methods

Reference number
Publication numbering
As from 1 January 1997 all IEC publications are issued with a designation in the
60000 series. For example, IEC 34-1 is now referred to as IEC 60034-1.
Consolidated editions
The IEC is now publishing consolidated versions of its publications. For example,
edition numbers 1.0, 1.1 and 1.2 refer, respectively, to the base publication, the
base publication incorporating amendment 1 and the base publication incorporating
amendments 1 and 2.
Further information on IEC publications
The technical content of IEC publications is kept under constant review by the IEC,
thus ensuring that the content reflects current technology. Information relating to
this publication, including its validity, is available in the IEC Catalogue of
publications (see below) in addition to new editions, amendments and corrigenda.
Information on the subjects under consideration and work in progress undertaken
by the technical committee which has prepared this publication, as well as the list
of publications issued, is also available from the following:
• IEC Web Site (www.iec.ch)
• Catalogue of IEC publications
The on-line catalogue on the IEC web site (http://www.iec.ch/searchpub/cur_fut.htm)
enables you to search by a variety of criteria including text searches, technical
committees and date of publication. On-line information is also available on
recently issued publications, withdrawn and replaced publications, as well as
corrigenda.
• IEC Just Published
This summary of recently issued publications (http://www.iec.ch/online_news/
justpub/jp_entry.htm) is also available by email. Please contact the Customer
Service Centre (see below) for further information.
• Customer Service Centre
If you have any questions regarding this publication or need further assistance,
please contact the Customer Service Centre:
Email: custserv@iec.ch
Tel: +41 22 919 02 11
Fax: +41 22 919 03 00
INTERNATIONAL IEC
STANDARD 61164
Second edition
2004-03
Reliability growth –
Statistical test and estimation methods
 IEC 2004  Copyright - all rights reserved
No part of this publication may be reproduced or utilized in any form or by any means, electronic or
mechanical, including photocopying and microfilm, without permission in writing from the publisher.
International Electrotechnical Commission, 3, rue de Varembé, PO Box 131, CH-1211 Geneva 20, Switzerland
Telephone: +41 22 919 02 11 Telefax: +41 22 919 03 00 E-mail: inmail@iec.ch Web: www.iec.ch
PRICE CODE
Commission Electrotechnique Internationale XA
International Electrotechnical Commission
Международная Электротехническая Комиссия
For price, see current catalogue

– 2 – 61164  IEC:2004(E)
CONTENTS
FOREWORD.4
INTRODUCTION.6

1 Scope.7
2 Normative references.7
3 Terms and definitions .7
4 Symbols.8
5 Reliability growth models in design and test .12
6 Reliability growth models used for systems/products in design phase.13
6.1 Modified power law model for planning of reliability growth in product design
phase .13
6.2 Modified Bayesian IBM-Rosner model for planning reliability growth in design
phase .16
7 Reliability growth planning a tracking in the product reliability growth testing.18
7.1 Continuous reliability growth models .18
7.2 Discrete reliability growth model.20
8 Use of the power law model in planning reliability improvement test programmes.23
9 Statistical test and estimation procedures for continuous power law model.23
9.1 Overview.23
9.2 Growth tests and parameter estimation .23
9.3 Goodness-of-fit tests.27
9.4 Confidence intervals on the shape parameter.29
9.5 Confidence intervals on current MTBF.31
9.6 Projection technique.33

Annex A (informative) Examples for planning and analytical models used in design
and test phase of product development.37
Annex B (informative) The power law reliability growth model – Background
information.50

Bibliography.55

Figure 1 – Planned improvement of the average failure rate or reliability .12
Figure A.1 – Planned and achieved reliability growth – Example.40
Figure A.2 – Planned reliability growth using Bayesian reliability growth model.41
Figure A.3 − Scatter diagram of expected and observed test times at failure based on
data of Table A.2 with power law model .48
Figure A.4 − Observed and estimated accumulated failures/accumulated test time
based on data of Table A.2 with power law model.49

Table 1 – Categories of reliability growth models with clause references .13
Table 2 − Critical values for Cramér-von Mises goodness-of-fit test at 10 % level of
significance.34
Table 3 − Two-sided 90 % confidence intervals for MTBF from Type I testing .35

61164  IEC:2004(E) – 3 –
Table 4 − Two-sided 90 % confidence intervals for MTBF from Type II testing .36
Table A.1 – Calculation of the planning model for reliability growth in design phase .39
Table A.2 − Complete data − All relevant failures and accumulated test times for
Type I test .46
Table A.3 − Grouped data for Example 3 derived from Table A.2 .46
Table A.4 − Complete data for projected estimates in Example 4 − All relevant failures
and accumulated test times .47
Table A.5 − Distinct types of Category B failures, from Table A.4, with failure times,
time of first occurrence, number observed and effectiveness factors.47

– 4 – 61164  IEC:2004(E)
INTERNATIONAL ELECTROTECHNICAL COMMISSION
____________
RELIABILITY GROWTH –
STATISTICAL TEST AND ESTIMATION METHODS

FOREWORD
1) The International Electrotechnical Commission (IEC) is a worldwide organization for standardization comprising
all national electrotechnical committees (IEC National Committees). The object of IEC is to promote
international co-operation on all questions concerning standardization in the electrical and electronic fields. To
this end and in addition to other activities, IEC publishes International Standards, Technical Specifications,
Technical Reports, Publicly Available Specifications (PAS) and Guides (hereafter referred to as “IEC
Publication(s)”). Their preparation is entrusted to technical committees; any IEC National Committee interested
in the subject dealt with may participate in this preparatory work. International, governmental and non-
governmental organizations liaising with the IEC also participate in this preparation. IEC collaborates closely
with the International Organization for Standardization (ISO) in accordance with conditions determined by
agreement between the two organizations.
2) The formal decisions or agreements of IEC on technical matters express, as nearly as possible, an international
consensus of opinion on the relevant subjects since each technical committee has representation from all
interested IEC National Committees.
3) IEC Publications have the form of recommendations for international use and are accepted by IEC National
Committees in that sense. While all reasonable efforts are made to ensure that the technical content of IEC
Publications is accurate, IEC cannot be held responsible for the way in which they are used or for any
misinterpretation by any end user.
4) In order to promote international uniformity, IEC National Committees undertake to apply IEC Publications
transparently to the maximum extent possible in their national and regional publications. Any divergence
between any IEC Publication and the corresponding national or regional publication shall be clearly indicated in
the latter.
5) IEC provides no marking procedure to indicate its approval and cannot be rendered responsible for any
equipment declared to be in conformity with an IEC Publication.
6) All users should ensure that they have the latest edition of this publication.
7) No liability shall attach to IEC or its directors, employees, servants or agents including individual experts and
members of its technical committees and IEC National Committees for any personal injury, property damage or
other damage of any nature whatsoever, whether direct or indirect, or for costs (including legal fees) and
expenses arising out of the publication, use of, or reliance upon, this IEC Publication or any other IEC
Publications.
8) Attention is drawn to the Normative references cited in this publication. Use of the referenced publications is
indispensable for the correct application of this publication.
9) Attention is drawn to the possibility that some of the elements of this IEC Publication may be the subject of
patent rights. IEC shall not be held responsible for identifying any or all such patent rights.
International Standard IEC 61164 has been prepared by IEC technical committee 56:
Dependability.
This second edition cancels and replaces the first edition, published in 1995, and constitutes
a technical revision.
The main changes with respect to the previous edition are listed below:
− addition of two statistical models for reliability growth planning and tracking in the product
design phase;
− statistical methods for the reliability growth programme in the design phase of IEC 61014;
− addition of the discrete reliability growth model for the test phase;
− addition of the fixed number of faults model for the test phase;
− clarification of the symbols used for various models;
− addition of real life examples for most of the statistical models;
− numerical correction of tables in the reliability growth test example.

61164  IEC:2004(E) – 5 –
This standard should be used in conjunction with IEC 61014.
The text of this standard is based on the following documents:
FDIS Report on voting
56/920/FDIS 56/939/RVD
Full information on the voting for the approval of this standard can be found in the report on
voting indicated in the above table.
This publication has been drafted in accordance with the ISO/IEC Directives, Part 2.
The committee has decided that the contents of this publication will remain unchanged until
2011. At this date, the publication will be
• reconfirmed;
• withdrawn;
• replaced by a revised edition, or
• amended.
– 6 – 61164  IEC:2004(E)
INTRODUCTION
This International Standard describes the power law reliability growth model and related
projection model and gives step-by-step directions for their use. There are several reliability
growth models available, the power law model being one of the most widely used. This
standard provides procedures to estimate some or all of the quantities listed in Clauses 4, 6
and 7 of IEC 61014.
Two types of input are required. The first one is for reliability growth planning through analysis
and design improvements in the design phase in terms of the design phase duration, initial
reliability, reliability goal, and planned design improvements, along with their expected
magnitude. The second input, for reliability growth in the project validation phase, is for a data
set of accumulated test times at which relevant failures occurred, or were observed, for a
single system, and the time of termination of the test, if different from the time of the final
failure. It is assumed that the collection of data as input for the model begins after the
completion of any preliminary tests, such as environmental stress screening, intended to
stabilize the product's initial failure intensity.
Model parameters estimated from previous test results may be used to plan and predict the
course of future reliability growth programmes, provided the conditions are similar.
Some of the procedures may require computer programs, but these are not unduly complex.
This standard presents algorithms for which computer programs should be easy to construct.

61164  IEC:2004(E) – 7 –
RELIABILITY GROWTH –
STATISTICAL TEST AND ESTIMATION METHODS

1 Scope
This International Standard gives models and numerical methods for reliability growth assess-
ments based on failure data, which were generated in a reliability improvement programme.
These procedures deal with growth, estimation, confidence intervals for product reliability and
goodness-of-fit tests.
2 Normative references
The following referenced documents are indispensable for the application of this document.
For dated references, only the edition cited applies. For undated references, the latest edition
of the referenced document (including any amendments) applies.
IEC 60050(191):1990, International Electrotechnical Vocabulary (IEV) − Chapter 191:
Dependability and quality of service
IEC 60300-3-5:2001, Dependability management – Part 3-5: Application guide – Reliability
test conditions and statistical test principles
IEC 60605-4, Equipment reliability testing − Part 4: Statistical procedures for exponential
distribution – Point estimates, confidence intervals, prediction intervals and tolerance intervals
IEC 60605-6, Equipment reliability testing − Part 6: Tests for the validity of the constant
failure rate or constant failure intensity assumptions
IEC 61014:2003, Programmes for reliability growth
3 Terms and definitions
For the purposes of this document, the terms and definitions of IEC 60050(191) and
IEC 61014, together with the following terms and definitions, apply.
3.1
reliability goal
desired level of reliability that the product should have at the end of the reliability growth
programme
3.2
initial reliability
reliability that is estimated for the product in earlier design stages before any potential failure
modes or their causes have been mitigated by the design improvement
3.3
reliability growth model for the design phase
mathematical model that takes into consideration potential design improvements, and their
magnitude to express mathematically reliability growth from start to finish during the design
period
– 8 – 61164  IEC:2004(E)
3.4
average product failure rate
average product failure rate calculated from its reliability as estimated for a predetermined
time period
NOTE The change in this failure rate as a function of time is a result of the modifications of the product design.
3.5
delayed modification
corrective modification, which is incorporated into the product at the end of a test
NOTE A delayed modification is not incorporated during the test.
3.6
improvement effectiveness factor
fraction by which the intensity of a systematic failure is reduced by means of corrective
modification
3.7
type I test
time-terminated test
reliability growth test which is terminated at a predetermined time, or test with data available
through a time which does not correspond to a failure
3.8
type II test
failure-terminated test
reliability growth test which is terminated upon the accumulation of a specified number of
failures, or test with data available through a time which corresponds to a failure
4 Symbols
For the purposes of this standard, the following symbols apply.
a) For 6.1, clauses A.1 and B.3:
product lifetime such as mission, warranty period or operational time
T
initial product reliability
R ()T
λ initial average failure rate of product in design period
a0
number of design modifications at any time during the design period
d()t
α reliability growth rate resultant from fault mitigation
D
D total number of implemented design improvements
total duration of the design period available for the design improvements
t
D
t
time variable during the design period from 0 to t
D
average failure rate of product as a function of time during the design period
λ ()t
a
λ ()t
goal average failure rate at the end of the design period t
aG D
D
61164  IEC:2004(E) – 9 –
reliability goal of the product to be attained during design period
R ()T
G
R()t,T reliability of product as a function of time and design improvements

b) For 6.2, clauses A.2 and B.4:
reliability goal of the product to be attained during design period
R ()T
G
total duration of the design period
t
D
α reliability growth rate during design period
D
rate of non-systematic (or residual) failures
λ
NS
D total number of predicted or implemented design improvements within design
period to address weaknesses
K total number of distinct classes of fault
j,k,i general purpose indicators
probability of j-th design weakness in fault class k resulting in failure during the
p
kj
specified life of the product
expected number of design weaknesses in fault class k resulting in failure during
η
k
the specified life of the product
total number of predicted or implemented design improvements within design
D
k
period to address faults in fault class k
failure rate of design weaknesses categorized in fault class k
λ
k
initial reliability at time T
R()T
I
reliability of product as a function of T
R()T
expected time to reach reliability goal
t
G
c) For 7.1.1, 7.1.2, Clauses 9, A.4, B.1, and B.2:
D total number of design modifications carried out during product design period to
mitigate identified faults
t total duration of the design period available for potential design modifications
D
t
time variable (during design period 0 ≤ t ≤ t )
D
d(t) number of design modifications at any given time t during design period from 0 to
t
D
reliability growth rate during the design period
α
D
initial average failure rate of a product in design
λ
a0
– 10 – 61164  IEC:2004(E)
product average failure rate variable as a function of time during the design period
λ (t)
a
(0 to t )
D
R (T) initial product reliability calculated for a time T (mission or other predetermined
time)
R (T) product reliability goal to be attained through design improvement, calculated for a
G
predetermined time
R(t) product reliability increase as a function of time and design improvements
goal average failure rate
λ
aG
T
predetermined time during a product life (mission, warranty, life)
λ scale parameter for the power law model
shape parameter for the power law model
β
CV critical value for hypothesis test
d number of intervals for grouped data analysis
mean and individual improvement effectiveness factors
E, E , E
i j
I
number of distinct types of category B failures observed
i, j
general purpose indices
K
number of category A failures
A
K number of category B failures
B
l
K
number of i-th type category B failures observed: K = K
i
B i
∑
i=1
M parameter of the Cramér-von Mises test (statistical)
N number of relevant failures
number of relevant failures in i-th interval
N
i
N(T) accumulated number of failures up to test time T
E[N(T)] expected accumulated number of failures up to test time T
t(i−1); t(i) endpoints of i-th interval of test time for grouped data analysis
T current accumulated relevant test time
accumulated relevant test time at the i-th failure
T
i
T total accumulated relevant test times for type II test
N
T* total accumulated relevant test times for type I test
2 2
() γ fractile of the χ distribution with ν degrees of freedom
χν
γ
z
general symbol for failure intensity
u
γ fractile of the standard normal distribution
γ
z projected failure intensity
p
z(T)
current failure intensity at time T (relevant test time)
θ(T) current instantaneous mean time between failures
θ projected mean time between failures
p
61164  IEC:2004(E) – 11 –
p probability of success at the stage i of product modification in design phase
j
described in Barlow, Proschan and Scheuer discrete reliability growth model
N(t) number of non-random type faults remaining at time t>0 in IBM/Rosner continuous

reliability growth model
g fraction that a product/equipment is debugged as given in the IBM/Rosner

reliability growth model
E exposure time of item
failure rate of non-systematic (residual) failures
λ
NS
λ failure rate of systematic failures
S
failure rate of the k-th failure class
μ
k
D number of potential design weakness in failure class k
k
p probability that the j-th potential design weakness associated with failure class k
k,j
will result in failure
t
expected design phase time to achieve goal reliability, R (T)
E G
t duration of design phase
D
R (T) initial reliability at time T
I
parameter to represent the expected growth rate
α
the expected number of design weaknesses associated with failure class k that will
λ
k
result in failure
K proportion of systematic (non-random) faults in product design at start of test
K number of faults in the product design at start of test
q fraction of faults removed through debugging on reliability growth test
q(T) fraction of original faults removed by time t
t
Expected time for removing fraction q of systematic faults in test
q
θ(T) Cumulative time between failures

d) Symbols used in the discrete reliability growth model, 7.2:
reliability, or success probability, of the i-th configuration
R
i
unreliability, or failure probability, of the i-th configuration
f = 1- R
i i
k number of stages and configurations
number of trials for stage i
n
i
number of failures for stage i
m
i
i
t = n
the cumulative number of trials through stage i
i j
∑
j =1
N( t ) the accumulated number of failures up through trial t
i i
E[]N(t ) the expected accumulated number of failure up through trial t
i i
λ,β scale and shape parameters for power law and discrete models

– 12 – 61164  IEC:2004(E)
5 Reliability growth models in design and test
The basic principles of reliability growth of a product are the same during design and test.
This is because both involve identifying and removing weaknesses to improve the product and
both measure that improvement by comparing the estimated reliability with the reliability goal.
The difference lies in the tools used to conduct design and test analysis and the models used
to measure reliability growth. IEC 61064 provides guidance on the construction of reliability
growth programmes and the analysis tools used in design and test. This standard provides
details about the models that can be used to measure reliability growth in different stages of
the product life cycle and for different types of items, such as repairable or one-shot items.
The mathematical models for reliability growth are constructed to estimate the growth
achieved and the projected reliability. Reliability growth models aim to support the planning of
reliability improvement programmes by estimating the number and the magnitude of the
changes during the design and development process or the test time required to reach a
specified reliability goal.
The reliability growth models can be formulated in terms of the failure rate (or intensity) or
probability of survival to a specified time (the reliability) as shown in Figure 1.

Initial λ
i
Goal R
f
Goal λ
f Initial R
i
12 3 4 5 12 3 4 5
Number of design improvements Number of design improvements
IEC  162/04
Figure 1 – Planned improvement of the average failure rate or reliability
Within this general framework many models for reliability growth exist. Table 1 provides a
summary of the main categories. As well as the distinction between design and test, the type
of data available will influence model selection. The continuous category refers to items that
operate through time, for example, repaired items. The discrete category refers to data that
are collected as if for a success/failure of a trial, for example, one-shot items. The procedures
used to estimate reliability growth are labelled classical or Bayesian. The former uses the
observed data only, while the latter uses both empirical data from design and test as well as
engineering knowledge, for example, regarding the anticipated number of failure modes of
concern.
Failure rate
Reliability (percent
survived)
61164  IEC:2004(E) – 13 –
Table 1 – Categories of reliability growth models with clause references
Time
Continuous Discrete
(time) (number of trials)
Classical 6.1 –
Model type
Design
Bayesian 6.2 –
Classical 7.1 7.2
Test
Bayesian – –
Many reliability models have been developed for analysing test data. This standard presents
one of the most popular growth models, the power law (also known as the AMSAA or the
Crow model) in both its continuous and discrete forms. This model is a generalization of the
Duane reliability growth model due to Crow [1] . Although Bayesian variants of these models
exist, they are not presented here. A review of the variety of reliability growth models
available for analysing test data can be found in Jewell [2, 3] and Xie [4].
There are fewer documented reports of reliability growth models being used in design.
Therefore a reliability growth planning model that is a modification of the power law for use in
design and a Bayesian variant of the IBM-Rosner model adapted for design have been
introduced. However, these are only given for products operating through continuous time.
In general, the choice of a reliability growth model involves a compromise between simplicity
and realism. Selection should be made according to the aforementioned criteria such as stage
of lifecycle and type of data, as well as by evaluating the validity of the assumptions
underpinning a specific model for the context to which it is to be applied. Further details about
the assumptions for the models described in this standard are given in Clauses 6 and 7. Note
that reliability growth models should not be regarded as infallible nor should they be applied
without discretion but used as statistical tools to aid engineering judgement.
6 Reliability growth models used for systems/products in design phase
6.1 Modified power law model for planning of reliability growth
in product design phase
6.1.1 General
The statistical procedure for the modified power law model for the planned reliability growth in
the product design phase concerns the necessary implementation of the design reliability
improvements by mitigation of a failure mode, or by reduction of its probability of occurrence,
and the time from the beginning of design to that improvement.
This model is used for planning purposes (and not for data analysis), to estimate the number
or the magnitude of improvements in the original design to increase its reliability from that
initially assessed to its goal value. The assumption of a power law for this model is justified by
the fact that the early improvements will be those that will contribute the most to the reliability
improvement, that is, the failure modes with the highest probability of occurrence will be
addressed first, followed by improvements of lesser and lesser reliability contribution. The
actual reliability values achieved in the course of the design are then plotted corresponding to
the design time when they were realized and compared to the model. This model is thus used
to plan the strategies necessary for reliability improvement of a design during the available
time period from the initial design revision until the design is completed and released for
production.
___________
Figures in square brackets refer to the bibliography.

– 14 – 61164  IEC:2004(E)
6.1.2 Planning model for the reliability growth during the product design period
The Krasich reliability growth planning model [5] is derived in the following manner. An
example of this model, as well as the spread sheet for its easy determination and plotting,
given initial and reliability goal of a product, are given in Annex A.
If the initial product reliability for the predetermined product operational life time T was
estimated by analysis or test to be R (T), then, assuming that its average failure rate is
constant, the initial average failure rate of that product corresponding to the time T is:
ln[]R (T )
λ = − (1)
a0
T
Assumption of applicability of the power law is justified by the fact that the faults which are
found to be the highest unreliability contributors are addressed first, and with design
modification (fault mitigation) the product failure rate is continuously improved with a function
d(t). The failure rate of the product design at any time during the design period is:
−α
D
λ (t) = λ ⋅[]1+ d(t) (2)
a a0
where
d(t) is the number of design modifications at any time during the design period;
α is the reliability growth rate resultant from fault mitigation;
D
D is the total number of implemented design improvements;
t is the total duration of the design period available for the design improvements.
D
With a linear approximation, the number of design modifications as a function of time can be
linearly distributed over the design period:
t
(3)
d(t) = D ⋅
t
D
The average failure rate as a function of time then becomes:
−α
D
 
t
  (4)
λ (t) = λ ⋅ 1 + D ⋅
a a0
 
t
 D 
If the goal product average failure rate given the product reliability goal is expressed as R (T),
G
then the goal average failure rate at the end of design period t , is approximated by:
D
−ln[R (T )]
G
(5)
λ (t ) =
aG D
T
At the same time:
−α
D
 
t
D −α
D
 
λ (t ) = λ ⋅ 1+ D ⋅ = λ ⋅()1+ D
aG D a0 a0
 
t
 D 
(6)
ln[]R (T )
−α
D
λ (t ) = − ⋅()1+ D
aG D
T
61164  IEC:2004(E) – 15 –
Substituting λ (t ) with the expression containing reliability goal and solving for D
aG D
[]ln R (T ) 
G
−ln 
 
ln[]R (T )
 
α
D
D = e − 1 (7)
Solving the same equation for the growth rate, expressed as a function of design
modifications and initial and reliability goal gives:
 
ln[]R (T)
ln 
 
ln[]R (T)
 G 
(8)
α =
D
ln(1+ D)
During the design period, continuous improvement of product reliability that it has to have at
the time T is a function of time t (the reliability growth model in the time period from 0 to t )
D
can be written as:
R(t,T) = exp()− λ (t) ⋅ T (9)
a
Substituting the expression for the average failure rate, the Krasich reliability growth model
for the design phase 0 < t < t , is derived as follows:
D
−α
 D 
 
t
 
(10)
R(t,T ) = exp − λ ⋅1+ D ⋅  ⋅ T
a0
 
 
 t 
 D 
 
−α
 D 
 
ln[]R (T) t
 
(11)
 
R(t,T) = exp ⋅ 1+ D ⋅ ⋅T
   
 T t 
D
 
 
In the above equation, expressing D in terms of initial and reliability goal, the reliability growth
as a function of time in design period, available for design improvements becomes:
−α
D
 1 
 
−
 
 
 ln[]R (T )  α
G
D
t +t⋅  −t
 
 D 
 
ln[]R (T )
 
  0  
   
 
 
t
D
 
 
 
 
 
R(t,T) = R (T) (12)
6.1.3 Tracking the achieved reliability growth
Tracking of the achieved reliability growth means a simple recalculation of the assessed
product reliability at the time the design was improved to account for the modifications. The
reliability value calculated for the same predetermined life or mission period is simply plotted
on the same plot with the reliability growth model for the corresponding design time.
The resultant entries to the graph can then be fitted with a best-fit line (power), or the values
on the graph may be simply connected with the straight lines and the resultant achieved
reliability is compared to the reliability growth model.

– 16 – 61164  IEC:2004(E)
Use of fault tree analysis with commercially available software makes assessment of the
reliability improvement easy and quick to accomplish and track as the product reliability is
automatically calculated based on the changes.
After completion of the product design and with the introduction of the product validation
phase, the planned reliability growth test may further improve product reliability or uncover
failure modes that were not accounted for during analytical evaluations. The final reliability
assessment of the completed design can then serve as the reliability goal for the reliability
growth testing.
An example of practical derivation and application of the planning growth model for reliability
improvement in design phase is shown in A.1.1. This real life example shows step by step
how the model is constructed and how it is used.
6.2 Modified Bayesian IBM-Rosner model for planning reliability growth in design
phase
6.2.1 General
A model is presented to describe the growth of reliability during the design phase of a
repairable item prepared by Quigley and Walls [6] to [8] and is based on a Bayesian
adaptation of the IBM-Rosner model [9] which was developed for analysing test data and is
described in 7.1.2.
It is assumed a design has been developed to a sufficient level of detail to provide an initial
estimate of reliability. It is further assumed that the reliability goal is specified. Modifications
to the design will be made with a view to improving reliability until the goal is achieved. The
model aims to capture the possible timings of the design modifications.
The model assumes that design review and re-assessments result in modifications with the
aim of improving reliability and achieving the goal. The rate of growth as measured by
advancing the initial reliability to the reliability goal is a function of the removal of aspects of
design that contribute to systematic failures. It is assumed through using the model under
consideration that there are greater improvements to reliability in the earlier re-design
compared with later stages of re-design.
The model can be used in two ways:
a) to predict the length of time required to achieve reliability goal by forecasting the reliability
of the design – this assumes that the expected growth rate can be quantified; or
b) to estimate the growth rate required to achieve the reliability goal from the initial estimate
during a specified design period duration – this assumes that the duration of the design
phase is fixed.
Details concerning the mathematical formulation of the model are given in Annex B.
6.2.2 Data requirements
Data are required concerning the reliability goal ( R ()T ) and either the duration of the design
G
phase ( t ) or the expected growth rate during design ( a ) according to the purpose of the
D D
model application.
The failure rate of non-systematic failures ( λ ) should be specified. This may be estimated
NS
from historical data for similar product designs operating in nominally identical environments.
All potential design improvements to mitigate the D potential weaknesses should be identified
and may be allocated to one of K fault classes as appropriate.

61164  IEC:2004(E) – 17 –
The probability of each design weakness within each fault class resulting in failure during the
specified life of the product should be estimated. This may be based on engineering
judgement. The probability for the j-th design weakness in fault class k is given by p .
kj
The expected number of design weaknesses in fault class k (η ) resulting in failure if no
k
modifications are implemented to the product design can be calculated using
D
k
η = p (13)
k ∑ kj
j=1
where D is the total number of design weaknesses expected in fault class k.
k
The failure rate for each fault class is required. These may be estimated from historical data
for similar product designs operating in nominally identical environments. The failure rate for
fault class k is given by λ .
k
6.2.3 Estimates of reliability growth and related parameters
In this subclause, equations are given to compute the key parameters of the reliability growth
model.
The initial reliability of the design at time T is calculated by
K
 
 
 
−λ T
k
R()T = exp − λ T + η (1− e ) (14)
 
I NS ∑ k
 
 
 k =1 
 
The reliability growth of the design at time T is given by
K
 
 
 
−α T −λ T
D k
R()T = exp − λ T + η e (1− e ) (15)
 
NS k
∑
 
 
k =1
 
 
If a growth rate is to be estimated, then the expected growth rate required to reach the
reliability goal, given the reliability goal and the specified duration of the design period, is
given by
K
 
−λ T
k
 η (1− e ) 
∑ k
 
k =1
ln
 
[]
− ln R ()T − λ T
G NS
 
 
 
α = (16)
D
t
D
If a growth rate has been specified, an estimate of the expected time to reach the reliability
goal is given by
K
 
−λ T
k
 η (1− e ) 
∑ k
 
k =1
ln
 
[]
− ln R ()T − λ T
G NS
 
 
 
t = (17)
G
α
D
– 18 – 61164  IEC:2004(E)
6.2.4 Tracking reliability growth during design phase
The tracking of the reliability growth with this modified Bayesian IBM-Rosner model is the
same as that described in 6.1.3.
7 Reliability growth planning a tracking in the product reliability growth
testing
7.1 Continuous reliability growth models
7.1.1 The power law model
The statistical procedures for the power law reliability growth model use the original relevant
failure and time data from the test. Except in the projection technique (see 9.6), the model is
applied to the complete set of relevant failures (as in IEC 61014, Figure 2 and Figure 4,
characteristic 3) without subdivision into categories.
The basic equations for the power law model are given in this subclause. Background
information on the model is given in Annex B.
The expected accumulated number of failures up to test time T is given by:
β
E[]N()T =λT , with λ > 0, β > 0, T > 0 (18)
where
λ is the scale parameter;
β is the shape parameter (a function of the general effectiveness of the improvements;
0 < β < 1, corresponds to reliability growth; β = 1 corresponds to no reliability growth; β > 1
corresponds to negative reliability growth).
The current failure intensity after T h of testing is given by:
d
β −1
z()T = E[]N()T =λβT , with T > 0 (19)
dT
Thus, parameters λ and β both affect the failure intensity achieved in a given time. The
equation represents in effect the slope of a tangent to the N(T) against T characteristic at time
T as shown in IEC 61014, Figure 9.
The current mean time between failures after T h of testing is given by:
θ()T = (20)
z()T
Methods are given in 9.1 and 9.2 for maximum likelihood estimation of the parameters λ
and β. Subclause 9.3 gives goodness-of-fit tests for the model, and 9.4 and 9.5 discuss
confidence interval procedures. An extension of the model for reliability growth projections is
given in 9.6.
The model has the following characteristic features:
− It is simple to evaluate.
61164  IEC:2004(E) – 19 –
− When the parameters have been estimated from past programmes, it is a convenient tool
for planning future programmes employing similar conditions of testing and equal
improvement effectiveness (see Clause 7, and IEC 61014, 6.4.1 to 6.4.7).
()
− It gives the unrealistic indications that zT =∞ at T = 0 and that growth can be unending,
that is z(T) tends to zero as T tends to infinity. However, these limitations do not generally
affect its practical use.
− It is relatively slow and insensitive in indicating growth immediately after a corrective
modification, and so may give a low (that is, pessimistic) estimate of the final θ(T), unless
projection is used (see 9.6).
− The normal evaluation method assumes the observed times to be exact times of failure,
but an alternative approach is possible for groups of failures within a known time period
(see 9.2.2).
7.1.2 The fixed number of faults model
This model, also known as the IBM/Rosner model [9], assumes the following:
− there are random (constant intensity function) failures occurring at a rate z; and
− there is a fixed, but unknown, number of non-random design, manufacturing and
workmanship faults present in the product at the beginning of testing.
The model limitation is the assumption that the effectiveness factor for fault mitigation is equal
to unity.
Rate of change of N(T), with respect to time, is proportional to the number of non-random
faults remaining at time T:
d N(T )
= −K ⋅ N(T )
dT
(21)
thus :
−K ⋅T +c
N(T ) = e
If the number of faults at T = 0 is K , then:
−K ⋅T
N (T ) = K ⋅ e (22)
with an assumption that
T > 0, K , K > 0
1 2
If E[N(T)] is to be the expected cumulative number of faults up to time T, then:
−K ⋅T
)
E[N(T )] = z ⋅ T + K ⋅ (1− e (23)
The equation above means that by the time T, the total number of faults is equal to the sum of
random and non-random faults. Here E[N(0)] = 0.
With time approaching infinity:
(24)
E[N (T )] → ∞
Since the model is non-linear, the estimate of λ, K , and K has to be accomplished by
1 2
iterative methods.
– 20 – 61164  IEC:2004(E)
The model allows prediction of the time when the product is "q" fraction debugged (fraction of
the original non-random faults removed, while 0 The number of non-random faults removed by the time t is:
−K ⋅T
N(0) − N(t) = K − K ⋅ e (25)
1 1
Hence the fraction of the original faults removed by time t is:
−K ⋅T
K − K ⋅ e
− K ⋅T
1 1
q(T) = = 1− e (26)
K
ˆ
Thus, having estimated K as ( K ), the time necessary for a desired q, of non-random faults
to have been removed is found as:
ln(1−
...